Last changed in Agentgateway v1.0.1
APIVersion defines the versioned schema of this representation of an object. Servers should convert recognized schemas to the latest internal value, and may reject unrecognized values. More info: https://git.k8s.io/community/contributors/devel/sig-architecture/api-conventions.md#resources
Kind is a string value representing the REST resource this object represents. Servers may infer this from the endpoint the client submits requests to. Cannot be updated. In CamelCase. More info: https://git.k8s.io/community/contributors/devel/sig-architecture/api-conventions.md#types-kinds
spec defines the desired state of AgentgatewayBackend.
ai represents a LLM backend.
groups specifies a list of groups in priority order where each group defines a set of LLM providers. The priority determines the priority of the backend endpoints chosen. Note: provider names must be unique across all providers in all priority groups. Backend policies may target a specific provider by name using targetRefs[].sectionName. Example configuration with two priority groups: ```yaml groups: - providers: - azureopenai: deploymentName: gpt-4o-mini apiVersion: 2024-02-15-preview endpoint: ai-gateway.openai.azure.com - providers: - azureopenai: deploymentName: gpt-4o-mini-2 apiVersion: 2024-02-15-preview endpoint: ai-gateway-2.openai.azure.com policies: auth: secretRef: name: azure-secret ```
providers specifies a list of LLM providers within this group. Each provider is treated equally in terms of priority, with automatic weighting based on health.
Anthropic provider
Optional: Override the model name, such as `gpt-4o-mini`. If unset, the model name is taken from the request.
Azure OpenAI provider
The version of the Azure OpenAI API to use. For more information, see the [Azure OpenAI API version reference](https://learn.microsoft.com/en-us/azure/ai-foundry/?view=foundry-classicreference#api-specs). If unset, defaults to "v1"
The name of the Azure OpenAI model deployment to use. For more information, see the [Azure OpenAI model docs](https://learn.microsoft.com/en-us/azure/ai-foundry/foundry-models/concepts/models-sold-directly-by-azure?view=foundry-classic). This is required if ApiVersion is not 'v1'. For v1, the model can be set in the request.
The endpoint for the Azure OpenAI API to use, such as `my-endpoint.openai.azure.com`. If the scheme is included, it is stripped.
Bedrock provider
Guardrail configures the Guardrail policy to use for the backend. See <https://docs.aws.amazon.com/bedrock/latest/userguide/guardrails.html> If not specified, the AWS Guardrail policy will not be used.
GuardrailIdentifier is the identifier of the Guardrail policy to use for the backend.
GuardrailVersion is the version of the Guardrail policy to use for the backend.
Optional: Override the model name, such as `gpt-4o-mini`. If unset, the model name is taken from the request.
Region is the AWS region to use for the backend. Defaults to us-east-1 if not specified.
Gemini provider
Optional: Override the model name, such as `gemini-2.5-pro`. If unset, the model name is taken from the request.
Host specifies the hostname to send the requests to. If not specified, the default hostname for the provider is used.
Name of the provider. Policies can target this provider by name.
OpenAI provider
Optional: Override the model name, such as `gpt-4o-mini`. If unset, the model name is taken from the request.
Path specifies the URL path to use for the LLM provider API requests. This is useful when you need to route requests to a different API endpoint while maintaining compatibility with the original provider's API structure. If not specified, the default path for the provider is used.
policies controls policies for communicating with this backend. Policies may also be set in AgentgatewayPolicy, or in the top level AgentgatewayBackend. policies are merged on a field-level basis, with order: AgentgatewayPolicy < AgentgatewayBackend < AgentgatewayBackend LLM provider (this field).
ai specifies settings for AI workloads. This is only applicable when connecting to a Backend of type 'ai'.
Provide defaults to merge with user input fields. If the field is already set, the field in the request is used.
The name of the field.
The field default value, which can be any JSON Data Type.
ModelAliases maps friendly model names to actual provider model names. Example: {"fast": "gpt-3.5-turbo", "smart": "gpt-4-turbo"} Note: This field is only applicable when using the agentgateway data plane.
Provide overrides to merge with user input fields. If the field is already set, the field will be overwritten.
The name of the field.
The field default value, which can be any JSON Data Type.
Enrich requests sent to the LLM provider by appending and prepending system prompts. This can be configured only for LLM providers that use the `CHAT` or `CHAT_STREAMING` API route type.
A list of messages to be appended to the prompt sent by the client.
String content of the message.
Role of the message. The available roles depend on the backend LLM provider model, such as `SYSTEM` or `USER` in the OpenAI API.
A list of messages to be prepended to the prompt sent by the client.
String content of the message.
Role of the message. The available roles depend on the backend LLM provider model, such as `SYSTEM` or `USER` in the OpenAI API.
promptCaching enables automatic prompt caching for supported providers (AWS Bedrock). Reduces API costs by caching static content like system prompts and tool definitions. Only applicable for Bedrock Claude 3+ and Nova models.
CacheMessages enables caching for conversation messages. Caches all messages in the conversation for cost savings.
CacheSystem enables caching for system prompts. Inserts a cache point after all system messages.
CacheTools enables caching for tool definitions. Inserts a cache point after all tool specifications.
MinTokens specifies the minimum estimated token count before caching is enabled. Uses rough heuristic (word count × 1.3) to estimate tokens. Bedrock requires at least 1,024 tokens for caching to be effective.
promptGuard enables adding guardrails to LLM requests and responses.
Prompt guards to apply to requests sent by the client.
bedrockGuardrails configures AWS Bedrock Guardrails for prompt guarding.
GuardrailIdentifier is the identifier of the Guardrail policy to use for the backend.
policies controls policies for communicating with AWS Bedrock Guardrails.
auth defines settings for managing authentication to the backend
Auth specifies an explicit AWS authentication method for the backend. When omitted, we will try to use the default AWS SDK authentication methods.
SecretRef references a Kubernetes Secret containing the AWS credentials. The Secret must have keys "accessKey", "secretKey", and optionally "sessionToken".
Name of the referent. This field is effectively required, but due to backwards compatibility is allowed to be empty. Instances of this type with an empty value here are almost certainly wrong. More info: https://kubernetes.io/docs/concepts/overview/working-with-objects/names/#names
Azure specifies an Azure authentication method for the backend.
Details for managed identity authentication
SecretRef references a Kubernetes Secret containing the Azure credentials. The Secret must have keys "clientId", "tenantId", and "clientSecret".
Name of the referent. This field is effectively required, but due to backwards compatibility is allowed to be empty. Instances of this type with an empty value here are almost certainly wrong. More info: https://kubernetes.io/docs/concepts/overview/working-with-objects/names/#names
Auth specifies to use a Google authentication method for the backend. When omitted, we will try to use the default AWS SDK authentication methods.
audience allows explicitly configuring the `aud` of the ID Token. Ony valid with `IdToken` type. If not set, the aud is automatically derived from the backend hostname.
The type of token to generate. To authenticate to GCP services, generally an AccessToken is used. To authenticate to CloudRun, an IdToken is used.
key provides an inline key to use as the value of the Authorization header. This option is the least secure; usage of a Secret is preferred.
passthrough passes through an existing token that has been sent by the client and validated. Other policies, like JWT and API Key authentication, will strip the original client credentials. Passthrough backend authentication causes the original token to be added back into the request. If there are no client authentication policies on the request, the original token would be unchanged, so this would have no effect.
secretRef references a Kubernetes secret storing the key to use the authorization value. This must be stored in the 'Authorization' key.
Name of the referent. This field is effectively required, but due to backwards compatibility is allowed to be empty. Instances of this type with an empty value here are almost certainly wrong. More info: https://kubernetes.io/docs/concepts/overview/working-with-objects/names/#names
health defines settings for passive and active health checking.
Eviction defines settings for evicting unhealthy backends.
ConsecutiveFailures is the number of consecutive unhealthy responses required before the backend is evicted. For example, a value of 5 means the backend must receive 5 unhealthy responses in a row before being evicted. When both consecutiveFailures and healthThreshold are set, the backend is evicted when either condition is met. When neither is set, a single unhealthy response can trigger eviction.
Duration specifies the base time a backend should be evicted after being marked unhealthy. Subsequent evictions use multiplicative backoff (duration * times_evicted). If all endpoints are evicted, the load balancer falls back to returning evicted endpoints rather than failing entirely. If unset, defaults to 3s.
HealthThreshold is the EWMA (exponentially-weighted moving average) health score threshold, expressed as 0–100. When set, a backend is only evicted if its computed health drops below this value after an unhealthy response. For example, 50 means the backend is evicted when its EWMA health falls below 50% following failures. Unlike consecutiveFailures (which counts consecutive failures), this uses a sliding-window average so a single success in a stream of failures can delay eviction. When both consecutiveFailures and healthThreshold are set, the backend is evicted when either condition is met. When neither is set, a single unhealthy response triggers eviction.
RestoreHealth is the health score (0–100) assigned to a backend when it returns from eviction. For gradual recovery, set below 100; for full recovery immediately, set 100. If unset, the backend resumes with the health it had when evicted.
UnhealthyCondition is a CEL expression that determines whether a response indicates an unhealthy backend. When the expression evaluates to true, the backend is considered unhealthy and may be evicted. For example, to evict on 5xx responses: `response.code >= 500`. When unset, any 5xx response, or a connection failure, is treated as unhealthy. This default lowers the backend's health score but does not trigger eviction on its own.
http defines settings for managing HTTP requests to the backend.
requestTimeout specifies the deadline for receiving a response from the backend.
version specifies the HTTP protocol version to use when connecting to the backend. If not specified, the version is automatically determined: * Service types can specify it with 'appProtocol' on the Service port. * If traffic is identified as gRPC, HTTP2 is used. * If the incoming traffic was plaintext HTTP, the original protocol will be used. * If the incoming traffic was HTTPS, HTTP1 will be used. This is because most clients will transparently upgrade HTTPS traffic to HTTP2, even if the backend doesn't support it
tcp defines settings for managing TCP connections to the backend.
connectTimeout defines the deadline for establishing a connection to the destination.
keepAlive defines settings for enabling TCP keepalives on the connection.
interval specifies the number of seconds between keep-alive probes. If unset, this defaults to 180s.
retries specifies the maximum number of keep-alive probes to send before dropping the connection. If unset, this defaults to 9.
time specifies the number of seconds a connection needs to be idle before keep-alive probes start being sent. If unset, this defaults to 180s.
tls defines settings for managing TLS connections to the backend. If this field is set, TLS will be initiated to the backend; the system trusted CA certificates will be used to validate the server, and the SNI will automatically be set based on the destination.
alpnProtocols sets the Application Level Protocol Negotiation (ALPN) value to use in the TLS handshake. If not present, defaults to ["h2", "http/1.1"].
caCertificateRefs defines the CA certificate ConfigMap to use to verify the server certificate. If unset, the system's trusted certificates are used.
Name of the referent. This field is effectively required, but due to backwards compatibility is allowed to be empty. Instances of this type with an empty value here are almost certainly wrong. More info: https://kubernetes.io/docs/concepts/overview/working-with-objects/names/#names
insecureSkipVerify originates TLS but skips verification of the backend's certificate. WARNING: This is an insecure option that should only be used if the risks are understood. There are two modes: * All disables all TLS verification * Hostname verifies the CA certificate is trusted, but ignores any mismatch of hostname/SANs. Note that this method is still insecure; prefer setting verifySubjectAltNames to customize the valid hostnames if possible.
mtlsCertificateRef enables mutual TLS to the backend, using the specified key (tls.key) and cert (tls.crt) from the refenced Secret. An optional 'ca.cert' field, if present, will be used to verify the server certificate if present. If caCertificateRefs is also specified, the caCertificateRefs field takes priority. If unspecified, no client certificate will be used.
Name of the referent. This field is effectively required, but due to backwards compatibility is allowed to be empty. Instances of this type with an empty value here are almost certainly wrong. More info: https://kubernetes.io/docs/concepts/overview/working-with-objects/names/#names
sni specifies the Server Name Indicator (SNI) to be used in the TLS handshake. If unset, the SNI is automatically set based on the destination hostname.
verifySubjectAltNames specifies the Subject Alternative Names (SAN) to verify in the server certificate. If not present, the destination hostname is automatically used.
transformation is used to mutate and transform requests and responses sent to and from the backend.
request is used to modify the request path.
add is a list of headers to add to the request and what that value should be set to. If there is already a header with these values then append the value as an extra entry.
the name of the header to add.
value is the CEL expression to apply to generate the output value for the header.
body controls manipulation of the HTTP body.
metadata stores CEL-evaluated values under the `metadata` CEL variable for subsequent policy evaluations. metadata is evaluated before header or body transformations.
Remove is a list of header names to remove from the request/response.
set is a list of headers and the value they should be set to.
the name of the header to add.
value is the CEL expression to apply to generate the output value for the header.
response is used to modify the response path.
add is a list of headers to add to the request and what that value should be set to. If there is already a header with these values then append the value as an extra entry.
the name of the header to add.
value is the CEL expression to apply to generate the output value for the header.
body controls manipulation of the HTTP body.
metadata stores CEL-evaluated values under the `metadata` CEL variable for subsequent policy evaluations. metadata is evaluated before header or body transformations.
Remove is a list of header names to remove from the request/response.
set is a list of headers and the value they should be set to.
the name of the header to add.
value is the CEL expression to apply to generate the output value for the header.
tunnel defines settings for managing tunnel connections (like HTTPS_PROXY) to the backend.
backendRef references the proxy server to reach. Supported types: Service and Backend.
Group is the group of the referent. For example, "gateway.networking.k8s.io". When unspecified or empty string, core API group is inferred.
Kind is the Kubernetes resource kind of the referent. For example "Service". Defaults to "Service" when not specified. ExternalName services can refer to CNAME DNS records that may live outside of the cluster and as such are difficult to reason about in terms of conformance. They also may not be safe to forward to (see CVE-2021-25740 for more information). Implementations SHOULD NOT support ExternalName Services. Support: Core (Services with a type other than ExternalName) Support: Implementation-specific (Services with type ExternalName)
Name is the name of the referent.
Namespace is the namespace of the backend. When unspecified, the local namespace is inferred. Note that when a namespace different than the local namespace is specified, a ReferenceGrant object is required in the referent namespace to allow that namespace's owner to accept the reference. See the ReferenceGrant documentation for details. Support: Core
Port specifies the destination port number to use for this resource. Port is required when the referent is a Kubernetes Service. In this case, the port number is the service port number, not the target port. For other resources, destination port might be derived from the referent resource or this field.
Region is the AWS region where the guardrail is deployed (e.g., "us-west-2").
GuardrailVersion is the version of the Guardrail policy to use for the backend.
googleModelArmor configures Google Model Armor for prompt guarding.
Location is the Google Cloud location (e.g., "us-central1"). Defaults to "us-central1" if not specified.
policies controls policies for communicating with Google Model Armor.
auth defines settings for managing authentication to the backend
Auth specifies an explicit AWS authentication method for the backend. When omitted, we will try to use the default AWS SDK authentication methods.
SecretRef references a Kubernetes Secret containing the AWS credentials. The Secret must have keys "accessKey", "secretKey", and optionally "sessionToken".
Name of the referent. This field is effectively required, but due to backwards compatibility is allowed to be empty. Instances of this type with an empty value here are almost certainly wrong. More info: https://kubernetes.io/docs/concepts/overview/working-with-objects/names/#names
Azure specifies an Azure authentication method for the backend.
Details for managed identity authentication
SecretRef references a Kubernetes Secret containing the Azure credentials. The Secret must have keys "clientId", "tenantId", and "clientSecret".
Name of the referent. This field is effectively required, but due to backwards compatibility is allowed to be empty. Instances of this type with an empty value here are almost certainly wrong. More info: https://kubernetes.io/docs/concepts/overview/working-with-objects/names/#names
Auth specifies to use a Google authentication method for the backend. When omitted, we will try to use the default AWS SDK authentication methods.
audience allows explicitly configuring the `aud` of the ID Token. Ony valid with `IdToken` type. If not set, the aud is automatically derived from the backend hostname.
The type of token to generate. To authenticate to GCP services, generally an AccessToken is used. To authenticate to CloudRun, an IdToken is used.
key provides an inline key to use as the value of the Authorization header. This option is the least secure; usage of a Secret is preferred.
passthrough passes through an existing token that has been sent by the client and validated. Other policies, like JWT and API Key authentication, will strip the original client credentials. Passthrough backend authentication causes the original token to be added back into the request. If there are no client authentication policies on the request, the original token would be unchanged, so this would have no effect.
secretRef references a Kubernetes secret storing the key to use the authorization value. This must be stored in the 'Authorization' key.
Name of the referent. This field is effectively required, but due to backwards compatibility is allowed to be empty. Instances of this type with an empty value here are almost certainly wrong. More info: https://kubernetes.io/docs/concepts/overview/working-with-objects/names/#names
health defines settings for passive and active health checking.
Eviction defines settings for evicting unhealthy backends.
ConsecutiveFailures is the number of consecutive unhealthy responses required before the backend is evicted. For example, a value of 5 means the backend must receive 5 unhealthy responses in a row before being evicted. When both consecutiveFailures and healthThreshold are set, the backend is evicted when either condition is met. When neither is set, a single unhealthy response can trigger eviction.
Duration specifies the base time a backend should be evicted after being marked unhealthy. Subsequent evictions use multiplicative backoff (duration * times_evicted). If all endpoints are evicted, the load balancer falls back to returning evicted endpoints rather than failing entirely. If unset, defaults to 3s.
HealthThreshold is the EWMA (exponentially-weighted moving average) health score threshold, expressed as 0–100. When set, a backend is only evicted if its computed health drops below this value after an unhealthy response. For example, 50 means the backend is evicted when its EWMA health falls below 50% following failures. Unlike consecutiveFailures (which counts consecutive failures), this uses a sliding-window average so a single success in a stream of failures can delay eviction. When both consecutiveFailures and healthThreshold are set, the backend is evicted when either condition is met. When neither is set, a single unhealthy response triggers eviction.
RestoreHealth is the health score (0–100) assigned to a backend when it returns from eviction. For gradual recovery, set below 100; for full recovery immediately, set 100. If unset, the backend resumes with the health it had when evicted.
UnhealthyCondition is a CEL expression that determines whether a response indicates an unhealthy backend. When the expression evaluates to true, the backend is considered unhealthy and may be evicted. For example, to evict on 5xx responses: `response.code >= 500`. When unset, any 5xx response, or a connection failure, is treated as unhealthy. This default lowers the backend's health score but does not trigger eviction on its own.
http defines settings for managing HTTP requests to the backend.
requestTimeout specifies the deadline for receiving a response from the backend.
version specifies the HTTP protocol version to use when connecting to the backend. If not specified, the version is automatically determined: * Service types can specify it with 'appProtocol' on the Service port. * If traffic is identified as gRPC, HTTP2 is used. * If the incoming traffic was plaintext HTTP, the original protocol will be used. * If the incoming traffic was HTTPS, HTTP1 will be used. This is because most clients will transparently upgrade HTTPS traffic to HTTP2, even if the backend doesn't support it
tcp defines settings for managing TCP connections to the backend.
connectTimeout defines the deadline for establishing a connection to the destination.
keepAlive defines settings for enabling TCP keepalives on the connection.
interval specifies the number of seconds between keep-alive probes. If unset, this defaults to 180s.
retries specifies the maximum number of keep-alive probes to send before dropping the connection. If unset, this defaults to 9.
time specifies the number of seconds a connection needs to be idle before keep-alive probes start being sent. If unset, this defaults to 180s.
tls defines settings for managing TLS connections to the backend. If this field is set, TLS will be initiated to the backend; the system trusted CA certificates will be used to validate the server, and the SNI will automatically be set based on the destination.
alpnProtocols sets the Application Level Protocol Negotiation (ALPN) value to use in the TLS handshake. If not present, defaults to ["h2", "http/1.1"].
caCertificateRefs defines the CA certificate ConfigMap to use to verify the server certificate. If unset, the system's trusted certificates are used.
Name of the referent. This field is effectively required, but due to backwards compatibility is allowed to be empty. Instances of this type with an empty value here are almost certainly wrong. More info: https://kubernetes.io/docs/concepts/overview/working-with-objects/names/#names
insecureSkipVerify originates TLS but skips verification of the backend's certificate. WARNING: This is an insecure option that should only be used if the risks are understood. There are two modes: * All disables all TLS verification * Hostname verifies the CA certificate is trusted, but ignores any mismatch of hostname/SANs. Note that this method is still insecure; prefer setting verifySubjectAltNames to customize the valid hostnames if possible.
mtlsCertificateRef enables mutual TLS to the backend, using the specified key (tls.key) and cert (tls.crt) from the refenced Secret. An optional 'ca.cert' field, if present, will be used to verify the server certificate if present. If caCertificateRefs is also specified, the caCertificateRefs field takes priority. If unspecified, no client certificate will be used.
Name of the referent. This field is effectively required, but due to backwards compatibility is allowed to be empty. Instances of this type with an empty value here are almost certainly wrong. More info: https://kubernetes.io/docs/concepts/overview/working-with-objects/names/#names
sni specifies the Server Name Indicator (SNI) to be used in the TLS handshake. If unset, the SNI is automatically set based on the destination hostname.
verifySubjectAltNames specifies the Subject Alternative Names (SAN) to verify in the server certificate. If not present, the destination hostname is automatically used.
transformation is used to mutate and transform requests and responses sent to and from the backend.
request is used to modify the request path.
add is a list of headers to add to the request and what that value should be set to. If there is already a header with these values then append the value as an extra entry.
the name of the header to add.
value is the CEL expression to apply to generate the output value for the header.
body controls manipulation of the HTTP body.
metadata stores CEL-evaluated values under the `metadata` CEL variable for subsequent policy evaluations. metadata is evaluated before header or body transformations.
Remove is a list of header names to remove from the request/response.
set is a list of headers and the value they should be set to.
the name of the header to add.
value is the CEL expression to apply to generate the output value for the header.
response is used to modify the response path.
add is a list of headers to add to the request and what that value should be set to. If there is already a header with these values then append the value as an extra entry.
the name of the header to add.
value is the CEL expression to apply to generate the output value for the header.
body controls manipulation of the HTTP body.
metadata stores CEL-evaluated values under the `metadata` CEL variable for subsequent policy evaluations. metadata is evaluated before header or body transformations.
Remove is a list of header names to remove from the request/response.
set is a list of headers and the value they should be set to.
the name of the header to add.
value is the CEL expression to apply to generate the output value for the header.
tunnel defines settings for managing tunnel connections (like HTTPS_PROXY) to the backend.
backendRef references the proxy server to reach. Supported types: Service and Backend.
Group is the group of the referent. For example, "gateway.networking.k8s.io". When unspecified or empty string, core API group is inferred.
Kind is the Kubernetes resource kind of the referent. For example "Service". Defaults to "Service" when not specified. ExternalName services can refer to CNAME DNS records that may live outside of the cluster and as such are difficult to reason about in terms of conformance. They also may not be safe to forward to (see CVE-2021-25740 for more information). Implementations SHOULD NOT support ExternalName Services. Support: Core (Services with a type other than ExternalName) Support: Implementation-specific (Services with type ExternalName)
Name is the name of the referent.
Namespace is the namespace of the backend. When unspecified, the local namespace is inferred. Note that when a namespace different than the local namespace is specified, a ReferenceGrant object is required in the referent namespace to allow that namespace's owner to accept the reference. See the ReferenceGrant documentation for details. Support: Core
Port specifies the destination port number to use for this resource. Port is required when the referent is a Kubernetes Service. In this case, the port number is the service port number, not the target port. For other resources, destination port might be derived from the referent resource or this field.
ProjectID is the Google Cloud project ID.
TemplateID is the template ID for Google Model Armor.
openAIModeration passes prompt data through the OpenAI Moderations endpoint. See https://developers.openai.com/api/reference/resources/moderations for more information.
model specifies the moderation model to use. For example, `omni-moderation`.
policies controls policies for communicating with OpenAI.
auth defines settings for managing authentication to the backend
Auth specifies an explicit AWS authentication method for the backend. When omitted, we will try to use the default AWS SDK authentication methods.
SecretRef references a Kubernetes Secret containing the AWS credentials. The Secret must have keys "accessKey", "secretKey", and optionally "sessionToken".
Name of the referent. This field is effectively required, but due to backwards compatibility is allowed to be empty. Instances of this type with an empty value here are almost certainly wrong. More info: https://kubernetes.io/docs/concepts/overview/working-with-objects/names/#names
Azure specifies an Azure authentication method for the backend.
Details for managed identity authentication
SecretRef references a Kubernetes Secret containing the Azure credentials. The Secret must have keys "clientId", "tenantId", and "clientSecret".
Name of the referent. This field is effectively required, but due to backwards compatibility is allowed to be empty. Instances of this type with an empty value here are almost certainly wrong. More info: https://kubernetes.io/docs/concepts/overview/working-with-objects/names/#names
Auth specifies to use a Google authentication method for the backend. When omitted, we will try to use the default AWS SDK authentication methods.
audience allows explicitly configuring the `aud` of the ID Token. Ony valid with `IdToken` type. If not set, the aud is automatically derived from the backend hostname.
The type of token to generate. To authenticate to GCP services, generally an AccessToken is used. To authenticate to CloudRun, an IdToken is used.
key provides an inline key to use as the value of the Authorization header. This option is the least secure; usage of a Secret is preferred.
passthrough passes through an existing token that has been sent by the client and validated. Other policies, like JWT and API Key authentication, will strip the original client credentials. Passthrough backend authentication causes the original token to be added back into the request. If there are no client authentication policies on the request, the original token would be unchanged, so this would have no effect.
secretRef references a Kubernetes secret storing the key to use the authorization value. This must be stored in the 'Authorization' key.
Name of the referent. This field is effectively required, but due to backwards compatibility is allowed to be empty. Instances of this type with an empty value here are almost certainly wrong. More info: https://kubernetes.io/docs/concepts/overview/working-with-objects/names/#names
health defines settings for passive and active health checking.
Eviction defines settings for evicting unhealthy backends.
ConsecutiveFailures is the number of consecutive unhealthy responses required before the backend is evicted. For example, a value of 5 means the backend must receive 5 unhealthy responses in a row before being evicted. When both consecutiveFailures and healthThreshold are set, the backend is evicted when either condition is met. When neither is set, a single unhealthy response can trigger eviction.
Duration specifies the base time a backend should be evicted after being marked unhealthy. Subsequent evictions use multiplicative backoff (duration * times_evicted). If all endpoints are evicted, the load balancer falls back to returning evicted endpoints rather than failing entirely. If unset, defaults to 3s.
HealthThreshold is the EWMA (exponentially-weighted moving average) health score threshold, expressed as 0–100. When set, a backend is only evicted if its computed health drops below this value after an unhealthy response. For example, 50 means the backend is evicted when its EWMA health falls below 50% following failures. Unlike consecutiveFailures (which counts consecutive failures), this uses a sliding-window average so a single success in a stream of failures can delay eviction. When both consecutiveFailures and healthThreshold are set, the backend is evicted when either condition is met. When neither is set, a single unhealthy response triggers eviction.
RestoreHealth is the health score (0–100) assigned to a backend when it returns from eviction. For gradual recovery, set below 100; for full recovery immediately, set 100. If unset, the backend resumes with the health it had when evicted.
UnhealthyCondition is a CEL expression that determines whether a response indicates an unhealthy backend. When the expression evaluates to true, the backend is considered unhealthy and may be evicted. For example, to evict on 5xx responses: `response.code >= 500`. When unset, any 5xx response, or a connection failure, is treated as unhealthy. This default lowers the backend's health score but does not trigger eviction on its own.
http defines settings for managing HTTP requests to the backend.
requestTimeout specifies the deadline for receiving a response from the backend.
version specifies the HTTP protocol version to use when connecting to the backend. If not specified, the version is automatically determined: * Service types can specify it with 'appProtocol' on the Service port. * If traffic is identified as gRPC, HTTP2 is used. * If the incoming traffic was plaintext HTTP, the original protocol will be used. * If the incoming traffic was HTTPS, HTTP1 will be used. This is because most clients will transparently upgrade HTTPS traffic to HTTP2, even if the backend doesn't support it
tcp defines settings for managing TCP connections to the backend.
connectTimeout defines the deadline for establishing a connection to the destination.
keepAlive defines settings for enabling TCP keepalives on the connection.
interval specifies the number of seconds between keep-alive probes. If unset, this defaults to 180s.
retries specifies the maximum number of keep-alive probes to send before dropping the connection. If unset, this defaults to 9.
time specifies the number of seconds a connection needs to be idle before keep-alive probes start being sent. If unset, this defaults to 180s.
tls defines settings for managing TLS connections to the backend. If this field is set, TLS will be initiated to the backend; the system trusted CA certificates will be used to validate the server, and the SNI will automatically be set based on the destination.
alpnProtocols sets the Application Level Protocol Negotiation (ALPN) value to use in the TLS handshake. If not present, defaults to ["h2", "http/1.1"].
caCertificateRefs defines the CA certificate ConfigMap to use to verify the server certificate. If unset, the system's trusted certificates are used.
Name of the referent. This field is effectively required, but due to backwards compatibility is allowed to be empty. Instances of this type with an empty value here are almost certainly wrong. More info: https://kubernetes.io/docs/concepts/overview/working-with-objects/names/#names
insecureSkipVerify originates TLS but skips verification of the backend's certificate. WARNING: This is an insecure option that should only be used if the risks are understood. There are two modes: * All disables all TLS verification * Hostname verifies the CA certificate is trusted, but ignores any mismatch of hostname/SANs. Note that this method is still insecure; prefer setting verifySubjectAltNames to customize the valid hostnames if possible.
mtlsCertificateRef enables mutual TLS to the backend, using the specified key (tls.key) and cert (tls.crt) from the refenced Secret. An optional 'ca.cert' field, if present, will be used to verify the server certificate if present. If caCertificateRefs is also specified, the caCertificateRefs field takes priority. If unspecified, no client certificate will be used.
Name of the referent. This field is effectively required, but due to backwards compatibility is allowed to be empty. Instances of this type with an empty value here are almost certainly wrong. More info: https://kubernetes.io/docs/concepts/overview/working-with-objects/names/#names
sni specifies the Server Name Indicator (SNI) to be used in the TLS handshake. If unset, the SNI is automatically set based on the destination hostname.
verifySubjectAltNames specifies the Subject Alternative Names (SAN) to verify in the server certificate. If not present, the destination hostname is automatically used.
transformation is used to mutate and transform requests and responses sent to and from the backend.
request is used to modify the request path.
add is a list of headers to add to the request and what that value should be set to. If there is already a header with these values then append the value as an extra entry.
the name of the header to add.
value is the CEL expression to apply to generate the output value for the header.
body controls manipulation of the HTTP body.
metadata stores CEL-evaluated values under the `metadata` CEL variable for subsequent policy evaluations. metadata is evaluated before header or body transformations.
Remove is a list of header names to remove from the request/response.
set is a list of headers and the value they should be set to.
the name of the header to add.
value is the CEL expression to apply to generate the output value for the header.
response is used to modify the response path.
add is a list of headers to add to the request and what that value should be set to. If there is already a header with these values then append the value as an extra entry.
the name of the header to add.
value is the CEL expression to apply to generate the output value for the header.
body controls manipulation of the HTTP body.
metadata stores CEL-evaluated values under the `metadata` CEL variable for subsequent policy evaluations. metadata is evaluated before header or body transformations.
Remove is a list of header names to remove from the request/response.
set is a list of headers and the value they should be set to.
the name of the header to add.
value is the CEL expression to apply to generate the output value for the header.
tunnel defines settings for managing tunnel connections (like HTTPS_PROXY) to the backend.
backendRef references the proxy server to reach. Supported types: Service and Backend.
Group is the group of the referent. For example, "gateway.networking.k8s.io". When unspecified or empty string, core API group is inferred.
Kind is the Kubernetes resource kind of the referent. For example "Service". Defaults to "Service" when not specified. ExternalName services can refer to CNAME DNS records that may live outside of the cluster and as such are difficult to reason about in terms of conformance. They also may not be safe to forward to (see CVE-2021-25740 for more information). Implementations SHOULD NOT support ExternalName Services. Support: Core (Services with a type other than ExternalName) Support: Implementation-specific (Services with type ExternalName)
Name is the name of the referent.
Namespace is the namespace of the backend. When unspecified, the local namespace is inferred. Note that when a namespace different than the local namespace is specified, a ReferenceGrant object is required in the referent namespace to allow that namespace's owner to accept the reference. See the ReferenceGrant documentation for details. Support: Core
Port specifies the destination port number to use for this resource. Port is required when the referent is a Kubernetes Service. In this case, the port number is the service port number, not the target port. For other resources, destination port might be derived from the referent resource or this field.
Regular expression (regex) matching for prompt guards and data masking.
The action to take if a regex pattern is matched in a request or response. This setting applies only to request matches. PromptguardResponse matches are always masked by default. Defaults to `Mask`.
A list of built-in regex patterns to match against the request or response. Matches and built-ins are additive.
A list of regex patterns to match against the request or response. Matches and built-ins are additive.
A custom response message to return to the client. If not specified, defaults to "The request was rejected due to inappropriate content".
A custom response message to return to the client. If not specified, defaults to "The request was rejected due to inappropriate content".
The status code to return to the client. Defaults to 403.
Configure a webhook to forward requests to for prompt guarding.
backendRef references the webhook server to reach. Supported types: Service and Backend.
Group is the group of the referent. For example, "gateway.networking.k8s.io". When unspecified or empty string, core API group is inferred.
Kind is the Kubernetes resource kind of the referent. For example "Service". Defaults to "Service" when not specified. ExternalName services can refer to CNAME DNS records that may live outside of the cluster and as such are difficult to reason about in terms of conformance. They also may not be safe to forward to (see CVE-2021-25740 for more information). Implementations SHOULD NOT support ExternalName Services. Support: Core (Services with a type other than ExternalName) Support: Implementation-specific (Services with type ExternalName)
Name is the name of the referent.
Namespace is the namespace of the backend. When unspecified, the local namespace is inferred. Note that when a namespace different than the local namespace is specified, a ReferenceGrant object is required in the referent namespace to allow that namespace's owner to accept the reference. See the ReferenceGrant documentation for details. Support: Core
Port specifies the destination port number to use for this resource. Port is required when the referent is a Kubernetes Service. In this case, the port number is the service port number, not the target port. For other resources, destination port might be derived from the referent resource or this field.
ForwardHeaderMatches defines a list of HTTP header matches that will be used to select the headers to forward to the webhook. Request headers are used when forwarding requests and response headers are used when forwarding responses. By default, no headers are forwarded.
Name is the name of the HTTP Header to be matched. Name matching MUST be case-insensitive. (See https://tools.ietf.org/html/rfc7230#section-3.2). If multiple entries specify equivalent header names, only the first entry with an equivalent name MUST be considered for a match. Subsequent entries with an equivalent header name MUST be ignored. Due to the case-insensitivity of header names, "foo" and "Foo" are considered equivalent. When a header is repeated in an HTTP request, it is implementation-specific behavior as to how this is represented. Generally, proxies should follow the guidance from the RFC: https://www.rfc-editor.org/rfc/rfc7230.html#section-3.2.2 regarding processing a repeated header, with special handling for "Set-Cookie".
Type specifies how to match against the value of the header. Support: Core (Exact) Support: Implementation-specific (RegularExpression) Since RegularExpression HeaderMatchType has implementation-specific conformance, implementations can support POSIX, PCRE or any other dialects of regular expressions. Please read the implementation's documentation to determine the supported dialect.
Value is the value of HTTP Header to be matched. <gateway:experimental:description> Must consist of printable US-ASCII characters, optionally separated by single tabs or spaces. See: https://tools.ietf.org/html/rfc7230#section-3.2 </gateway:experimental:description> <gateway:experimental:validation:Pattern=`^[!-~]+([\t ]?[!-~]+)*$`>
Prompt guards to apply to responses returned by the LLM provider.
bedrockGuardrails configures AWS Bedrock Guardrails for prompt guarding.
GuardrailIdentifier is the identifier of the Guardrail policy to use for the backend.
policies controls policies for communicating with AWS Bedrock Guardrails.
auth defines settings for managing authentication to the backend
Auth specifies an explicit AWS authentication method for the backend. When omitted, we will try to use the default AWS SDK authentication methods.
SecretRef references a Kubernetes Secret containing the AWS credentials. The Secret must have keys "accessKey", "secretKey", and optionally "sessionToken".
Name of the referent. This field is effectively required, but due to backwards compatibility is allowed to be empty. Instances of this type with an empty value here are almost certainly wrong. More info: https://kubernetes.io/docs/concepts/overview/working-with-objects/names/#names
Azure specifies an Azure authentication method for the backend.
Details for managed identity authentication
SecretRef references a Kubernetes Secret containing the Azure credentials. The Secret must have keys "clientId", "tenantId", and "clientSecret".
Name of the referent. This field is effectively required, but due to backwards compatibility is allowed to be empty. Instances of this type with an empty value here are almost certainly wrong. More info: https://kubernetes.io/docs/concepts/overview/working-with-objects/names/#names
Auth specifies to use a Google authentication method for the backend. When omitted, we will try to use the default AWS SDK authentication methods.
audience allows explicitly configuring the `aud` of the ID Token. Ony valid with `IdToken` type. If not set, the aud is automatically derived from the backend hostname.
The type of token to generate. To authenticate to GCP services, generally an AccessToken is used. To authenticate to CloudRun, an IdToken is used.
key provides an inline key to use as the value of the Authorization header. This option is the least secure; usage of a Secret is preferred.
passthrough passes through an existing token that has been sent by the client and validated. Other policies, like JWT and API Key authentication, will strip the original client credentials. Passthrough backend authentication causes the original token to be added back into the request. If there are no client authentication policies on the request, the original token would be unchanged, so this would have no effect.
secretRef references a Kubernetes secret storing the key to use the authorization value. This must be stored in the 'Authorization' key.
Name of the referent. This field is effectively required, but due to backwards compatibility is allowed to be empty. Instances of this type with an empty value here are almost certainly wrong. More info: https://kubernetes.io/docs/concepts/overview/working-with-objects/names/#names
health defines settings for passive and active health checking.
Eviction defines settings for evicting unhealthy backends.
ConsecutiveFailures is the number of consecutive unhealthy responses required before the backend is evicted. For example, a value of 5 means the backend must receive 5 unhealthy responses in a row before being evicted. When both consecutiveFailures and healthThreshold are set, the backend is evicted when either condition is met. When neither is set, a single unhealthy response can trigger eviction.
Duration specifies the base time a backend should be evicted after being marked unhealthy. Subsequent evictions use multiplicative backoff (duration * times_evicted). If all endpoints are evicted, the load balancer falls back to returning evicted endpoints rather than failing entirely. If unset, defaults to 3s.
HealthThreshold is the EWMA (exponentially-weighted moving average) health score threshold, expressed as 0–100. When set, a backend is only evicted if its computed health drops below this value after an unhealthy response. For example, 50 means the backend is evicted when its EWMA health falls below 50% following failures. Unlike consecutiveFailures (which counts consecutive failures), this uses a sliding-window average so a single success in a stream of failures can delay eviction. When both consecutiveFailures and healthThreshold are set, the backend is evicted when either condition is met. When neither is set, a single unhealthy response triggers eviction.
RestoreHealth is the health score (0–100) assigned to a backend when it returns from eviction. For gradual recovery, set below 100; for full recovery immediately, set 100. If unset, the backend resumes with the health it had when evicted.
UnhealthyCondition is a CEL expression that determines whether a response indicates an unhealthy backend. When the expression evaluates to true, the backend is considered unhealthy and may be evicted. For example, to evict on 5xx responses: `response.code >= 500`. When unset, any 5xx response, or a connection failure, is treated as unhealthy. This default lowers the backend's health score but does not trigger eviction on its own.
http defines settings for managing HTTP requests to the backend.
requestTimeout specifies the deadline for receiving a response from the backend.
version specifies the HTTP protocol version to use when connecting to the backend. If not specified, the version is automatically determined: * Service types can specify it with 'appProtocol' on the Service port. * If traffic is identified as gRPC, HTTP2 is used. * If the incoming traffic was plaintext HTTP, the original protocol will be used. * If the incoming traffic was HTTPS, HTTP1 will be used. This is because most clients will transparently upgrade HTTPS traffic to HTTP2, even if the backend doesn't support it
tcp defines settings for managing TCP connections to the backend.
connectTimeout defines the deadline for establishing a connection to the destination.
keepAlive defines settings for enabling TCP keepalives on the connection.
interval specifies the number of seconds between keep-alive probes. If unset, this defaults to 180s.
retries specifies the maximum number of keep-alive probes to send before dropping the connection. If unset, this defaults to 9.
time specifies the number of seconds a connection needs to be idle before keep-alive probes start being sent. If unset, this defaults to 180s.
tls defines settings for managing TLS connections to the backend. If this field is set, TLS will be initiated to the backend; the system trusted CA certificates will be used to validate the server, and the SNI will automatically be set based on the destination.
alpnProtocols sets the Application Level Protocol Negotiation (ALPN) value to use in the TLS handshake. If not present, defaults to ["h2", "http/1.1"].
caCertificateRefs defines the CA certificate ConfigMap to use to verify the server certificate. If unset, the system's trusted certificates are used.
Name of the referent. This field is effectively required, but due to backwards compatibility is allowed to be empty. Instances of this type with an empty value here are almost certainly wrong. More info: https://kubernetes.io/docs/concepts/overview/working-with-objects/names/#names
insecureSkipVerify originates TLS but skips verification of the backend's certificate. WARNING: This is an insecure option that should only be used if the risks are understood. There are two modes: * All disables all TLS verification * Hostname verifies the CA certificate is trusted, but ignores any mismatch of hostname/SANs. Note that this method is still insecure; prefer setting verifySubjectAltNames to customize the valid hostnames if possible.
mtlsCertificateRef enables mutual TLS to the backend, using the specified key (tls.key) and cert (tls.crt) from the refenced Secret. An optional 'ca.cert' field, if present, will be used to verify the server certificate if present. If caCertificateRefs is also specified, the caCertificateRefs field takes priority. If unspecified, no client certificate will be used.
Name of the referent. This field is effectively required, but due to backwards compatibility is allowed to be empty. Instances of this type with an empty value here are almost certainly wrong. More info: https://kubernetes.io/docs/concepts/overview/working-with-objects/names/#names
sni specifies the Server Name Indicator (SNI) to be used in the TLS handshake. If unset, the SNI is automatically set based on the destination hostname.
verifySubjectAltNames specifies the Subject Alternative Names (SAN) to verify in the server certificate. If not present, the destination hostname is automatically used.
transformation is used to mutate and transform requests and responses sent to and from the backend.
request is used to modify the request path.
add is a list of headers to add to the request and what that value should be set to. If there is already a header with these values then append the value as an extra entry.
the name of the header to add.
value is the CEL expression to apply to generate the output value for the header.
body controls manipulation of the HTTP body.
metadata stores CEL-evaluated values under the `metadata` CEL variable for subsequent policy evaluations. metadata is evaluated before header or body transformations.
Remove is a list of header names to remove from the request/response.
set is a list of headers and the value they should be set to.
the name of the header to add.
value is the CEL expression to apply to generate the output value for the header.
response is used to modify the response path.
add is a list of headers to add to the request and what that value should be set to. If there is already a header with these values then append the value as an extra entry.
the name of the header to add.
value is the CEL expression to apply to generate the output value for the header.
body controls manipulation of the HTTP body.
metadata stores CEL-evaluated values under the `metadata` CEL variable for subsequent policy evaluations. metadata is evaluated before header or body transformations.
Remove is a list of header names to remove from the request/response.
set is a list of headers and the value they should be set to.
the name of the header to add.
value is the CEL expression to apply to generate the output value for the header.
tunnel defines settings for managing tunnel connections (like HTTPS_PROXY) to the backend.
backendRef references the proxy server to reach. Supported types: Service and Backend.
Group is the group of the referent. For example, "gateway.networking.k8s.io". When unspecified or empty string, core API group is inferred.
Kind is the Kubernetes resource kind of the referent. For example "Service". Defaults to "Service" when not specified. ExternalName services can refer to CNAME DNS records that may live outside of the cluster and as such are difficult to reason about in terms of conformance. They also may not be safe to forward to (see CVE-2021-25740 for more information). Implementations SHOULD NOT support ExternalName Services. Support: Core (Services with a type other than ExternalName) Support: Implementation-specific (Services with type ExternalName)
Name is the name of the referent.
Namespace is the namespace of the backend. When unspecified, the local namespace is inferred. Note that when a namespace different than the local namespace is specified, a ReferenceGrant object is required in the referent namespace to allow that namespace's owner to accept the reference. See the ReferenceGrant documentation for details. Support: Core
Port specifies the destination port number to use for this resource. Port is required when the referent is a Kubernetes Service. In this case, the port number is the service port number, not the target port. For other resources, destination port might be derived from the referent resource or this field.
Region is the AWS region where the guardrail is deployed (e.g., "us-west-2").
GuardrailVersion is the version of the Guardrail policy to use for the backend.
googleModelArmor configures Google Model Armor for prompt guarding.
Location is the Google Cloud location (e.g., "us-central1"). Defaults to "us-central1" if not specified.
policies controls policies for communicating with Google Model Armor.
auth defines settings for managing authentication to the backend
Auth specifies an explicit AWS authentication method for the backend. When omitted, we will try to use the default AWS SDK authentication methods.
SecretRef references a Kubernetes Secret containing the AWS credentials. The Secret must have keys "accessKey", "secretKey", and optionally "sessionToken".
Name of the referent. This field is effectively required, but due to backwards compatibility is allowed to be empty. Instances of this type with an empty value here are almost certainly wrong. More info: https://kubernetes.io/docs/concepts/overview/working-with-objects/names/#names
Azure specifies an Azure authentication method for the backend.
Details for managed identity authentication
SecretRef references a Kubernetes Secret containing the Azure credentials. The Secret must have keys "clientId", "tenantId", and "clientSecret".
Name of the referent. This field is effectively required, but due to backwards compatibility is allowed to be empty. Instances of this type with an empty value here are almost certainly wrong. More info: https://kubernetes.io/docs/concepts/overview/working-with-objects/names/#names
Auth specifies to use a Google authentication method for the backend. When omitted, we will try to use the default AWS SDK authentication methods.
audience allows explicitly configuring the `aud` of the ID Token. Ony valid with `IdToken` type. If not set, the aud is automatically derived from the backend hostname.
The type of token to generate. To authenticate to GCP services, generally an AccessToken is used. To authenticate to CloudRun, an IdToken is used.
key provides an inline key to use as the value of the Authorization header. This option is the least secure; usage of a Secret is preferred.
passthrough passes through an existing token that has been sent by the client and validated. Other policies, like JWT and API Key authentication, will strip the original client credentials. Passthrough backend authentication causes the original token to be added back into the request. If there are no client authentication policies on the request, the original token would be unchanged, so this would have no effect.
secretRef references a Kubernetes secret storing the key to use the authorization value. This must be stored in the 'Authorization' key.
Name of the referent. This field is effectively required, but due to backwards compatibility is allowed to be empty. Instances of this type with an empty value here are almost certainly wrong. More info: https://kubernetes.io/docs/concepts/overview/working-with-objects/names/#names
health defines settings for passive and active health checking.
Eviction defines settings for evicting unhealthy backends.
ConsecutiveFailures is the number of consecutive unhealthy responses required before the backend is evicted. For example, a value of 5 means the backend must receive 5 unhealthy responses in a row before being evicted. When both consecutiveFailures and healthThreshold are set, the backend is evicted when either condition is met. When neither is set, a single unhealthy response can trigger eviction.
Duration specifies the base time a backend should be evicted after being marked unhealthy. Subsequent evictions use multiplicative backoff (duration * times_evicted). If all endpoints are evicted, the load balancer falls back to returning evicted endpoints rather than failing entirely. If unset, defaults to 3s.
HealthThreshold is the EWMA (exponentially-weighted moving average) health score threshold, expressed as 0–100. When set, a backend is only evicted if its computed health drops below this value after an unhealthy response. For example, 50 means the backend is evicted when its EWMA health falls below 50% following failures. Unlike consecutiveFailures (which counts consecutive failures), this uses a sliding-window average so a single success in a stream of failures can delay eviction. When both consecutiveFailures and healthThreshold are set, the backend is evicted when either condition is met. When neither is set, a single unhealthy response triggers eviction.
RestoreHealth is the health score (0–100) assigned to a backend when it returns from eviction. For gradual recovery, set below 100; for full recovery immediately, set 100. If unset, the backend resumes with the health it had when evicted.
UnhealthyCondition is a CEL expression that determines whether a response indicates an unhealthy backend. When the expression evaluates to true, the backend is considered unhealthy and may be evicted. For example, to evict on 5xx responses: `response.code >= 500`. When unset, any 5xx response, or a connection failure, is treated as unhealthy. This default lowers the backend's health score but does not trigger eviction on its own.
http defines settings for managing HTTP requests to the backend.
requestTimeout specifies the deadline for receiving a response from the backend.
version specifies the HTTP protocol version to use when connecting to the backend. If not specified, the version is automatically determined: * Service types can specify it with 'appProtocol' on the Service port. * If traffic is identified as gRPC, HTTP2 is used. * If the incoming traffic was plaintext HTTP, the original protocol will be used. * If the incoming traffic was HTTPS, HTTP1 will be used. This is because most clients will transparently upgrade HTTPS traffic to HTTP2, even if the backend doesn't support it
tcp defines settings for managing TCP connections to the backend.
connectTimeout defines the deadline for establishing a connection to the destination.
keepAlive defines settings for enabling TCP keepalives on the connection.
interval specifies the number of seconds between keep-alive probes. If unset, this defaults to 180s.
retries specifies the maximum number of keep-alive probes to send before dropping the connection. If unset, this defaults to 9.
time specifies the number of seconds a connection needs to be idle before keep-alive probes start being sent. If unset, this defaults to 180s.
tls defines settings for managing TLS connections to the backend. If this field is set, TLS will be initiated to the backend; the system trusted CA certificates will be used to validate the server, and the SNI will automatically be set based on the destination.
alpnProtocols sets the Application Level Protocol Negotiation (ALPN) value to use in the TLS handshake. If not present, defaults to ["h2", "http/1.1"].
caCertificateRefs defines the CA certificate ConfigMap to use to verify the server certificate. If unset, the system's trusted certificates are used.
Name of the referent. This field is effectively required, but due to backwards compatibility is allowed to be empty. Instances of this type with an empty value here are almost certainly wrong. More info: https://kubernetes.io/docs/concepts/overview/working-with-objects/names/#names
insecureSkipVerify originates TLS but skips verification of the backend's certificate. WARNING: This is an insecure option that should only be used if the risks are understood. There are two modes: * All disables all TLS verification * Hostname verifies the CA certificate is trusted, but ignores any mismatch of hostname/SANs. Note that this method is still insecure; prefer setting verifySubjectAltNames to customize the valid hostnames if possible.
mtlsCertificateRef enables mutual TLS to the backend, using the specified key (tls.key) and cert (tls.crt) from the refenced Secret. An optional 'ca.cert' field, if present, will be used to verify the server certificate if present. If caCertificateRefs is also specified, the caCertificateRefs field takes priority. If unspecified, no client certificate will be used.
Name of the referent. This field is effectively required, but due to backwards compatibility is allowed to be empty. Instances of this type with an empty value here are almost certainly wrong. More info: https://kubernetes.io/docs/concepts/overview/working-with-objects/names/#names
sni specifies the Server Name Indicator (SNI) to be used in the TLS handshake. If unset, the SNI is automatically set based on the destination hostname.
verifySubjectAltNames specifies the Subject Alternative Names (SAN) to verify in the server certificate. If not present, the destination hostname is automatically used.
transformation is used to mutate and transform requests and responses sent to and from the backend.
request is used to modify the request path.
add is a list of headers to add to the request and what that value should be set to. If there is already a header with these values then append the value as an extra entry.
the name of the header to add.
value is the CEL expression to apply to generate the output value for the header.
body controls manipulation of the HTTP body.
metadata stores CEL-evaluated values under the `metadata` CEL variable for subsequent policy evaluations. metadata is evaluated before header or body transformations.
Remove is a list of header names to remove from the request/response.
set is a list of headers and the value they should be set to.
the name of the header to add.
value is the CEL expression to apply to generate the output value for the header.
response is used to modify the response path.
add is a list of headers to add to the request and what that value should be set to. If there is already a header with these values then append the value as an extra entry.
the name of the header to add.
value is the CEL expression to apply to generate the output value for the header.
body controls manipulation of the HTTP body.
metadata stores CEL-evaluated values under the `metadata` CEL variable for subsequent policy evaluations. metadata is evaluated before header or body transformations.
Remove is a list of header names to remove from the request/response.
set is a list of headers and the value they should be set to.
the name of the header to add.
value is the CEL expression to apply to generate the output value for the header.
tunnel defines settings for managing tunnel connections (like HTTPS_PROXY) to the backend.
backendRef references the proxy server to reach. Supported types: Service and Backend.
Group is the group of the referent. For example, "gateway.networking.k8s.io". When unspecified or empty string, core API group is inferred.
Kind is the Kubernetes resource kind of the referent. For example "Service". Defaults to "Service" when not specified. ExternalName services can refer to CNAME DNS records that may live outside of the cluster and as such are difficult to reason about in terms of conformance. They also may not be safe to forward to (see CVE-2021-25740 for more information). Implementations SHOULD NOT support ExternalName Services. Support: Core (Services with a type other than ExternalName) Support: Implementation-specific (Services with type ExternalName)
Name is the name of the referent.
Namespace is the namespace of the backend. When unspecified, the local namespace is inferred. Note that when a namespace different than the local namespace is specified, a ReferenceGrant object is required in the referent namespace to allow that namespace's owner to accept the reference. See the ReferenceGrant documentation for details. Support: Core
Port specifies the destination port number to use for this resource. Port is required when the referent is a Kubernetes Service. In this case, the port number is the service port number, not the target port. For other resources, destination port might be derived from the referent resource or this field.
ProjectID is the Google Cloud project ID.
TemplateID is the template ID for Google Model Armor.
Regular expression (regex) matching for prompt guards and data masking.
The action to take if a regex pattern is matched in a request or response. This setting applies only to request matches. PromptguardResponse matches are always masked by default. Defaults to `Mask`.
A list of built-in regex patterns to match against the request or response. Matches and built-ins are additive.
A list of regex patterns to match against the request or response. Matches and built-ins are additive.
A custom response message to return to the client. If not specified, defaults to "The response was rejected due to inappropriate content".
A custom response message to return to the client. If not specified, defaults to "The request was rejected due to inappropriate content".
The status code to return to the client. Defaults to 403.
Configure a webhook to forward responses to for prompt guarding.
backendRef references the webhook server to reach. Supported types: Service and Backend.
Group is the group of the referent. For example, "gateway.networking.k8s.io". When unspecified or empty string, core API group is inferred.
Kind is the Kubernetes resource kind of the referent. For example "Service". Defaults to "Service" when not specified. ExternalName services can refer to CNAME DNS records that may live outside of the cluster and as such are difficult to reason about in terms of conformance. They also may not be safe to forward to (see CVE-2021-25740 for more information). Implementations SHOULD NOT support ExternalName Services. Support: Core (Services with a type other than ExternalName) Support: Implementation-specific (Services with type ExternalName)
Name is the name of the referent.
Namespace is the namespace of the backend. When unspecified, the local namespace is inferred. Note that when a namespace different than the local namespace is specified, a ReferenceGrant object is required in the referent namespace to allow that namespace's owner to accept the reference. See the ReferenceGrant documentation for details. Support: Core
Port specifies the destination port number to use for this resource. Port is required when the referent is a Kubernetes Service. In this case, the port number is the service port number, not the target port. For other resources, destination port might be derived from the referent resource or this field.
ForwardHeaderMatches defines a list of HTTP header matches that will be used to select the headers to forward to the webhook. Request headers are used when forwarding requests and response headers are used when forwarding responses. By default, no headers are forwarded.
Name is the name of the HTTP Header to be matched. Name matching MUST be case-insensitive. (See https://tools.ietf.org/html/rfc7230#section-3.2). If multiple entries specify equivalent header names, only the first entry with an equivalent name MUST be considered for a match. Subsequent entries with an equivalent header name MUST be ignored. Due to the case-insensitivity of header names, "foo" and "Foo" are considered equivalent. When a header is repeated in an HTTP request, it is implementation-specific behavior as to how this is represented. Generally, proxies should follow the guidance from the RFC: https://www.rfc-editor.org/rfc/rfc7230.html#section-3.2.2 regarding processing a repeated header, with special handling for "Set-Cookie".
Type specifies how to match against the value of the header. Support: Core (Exact) Support: Implementation-specific (RegularExpression) Since RegularExpression HeaderMatchType has implementation-specific conformance, implementations can support POSIX, PCRE or any other dialects of regular expressions. Please read the implementation's documentation to determine the supported dialect.
Value is the value of HTTP Header to be matched. <gateway:experimental:description> Must consist of printable US-ASCII characters, optionally separated by single tabs or spaces. See: https://tools.ietf.org/html/rfc7230#section-3.2 </gateway:experimental:description> <gateway:experimental:validation:Pattern=`^[!-~]+([\t ]?[!-~]+)*$`>
routes defines how to identify the type of traffic to handle. The keys are URL path suffixes matched using ends-with comparison (e.g., "/v1/chat/completions"). The special "*" wildcard matches any path. If not specified, all traffic defaults to "completions" type.
Provide CEL transformations to compute and set fields in the request body. The expression result overwrites any existing value for that field. This has a higher priority than 'overrides' if both are set for the same key.
CEL expression used to compute the field value.
The name of the field to set.
auth defines settings for managing authentication to the backend
Auth specifies an explicit AWS authentication method for the backend. When omitted, we will try to use the default AWS SDK authentication methods.
SecretRef references a Kubernetes Secret containing the AWS credentials. The Secret must have keys "accessKey", "secretKey", and optionally "sessionToken".
Name of the referent. This field is effectively required, but due to backwards compatibility is allowed to be empty. Instances of this type with an empty value here are almost certainly wrong. More info: https://kubernetes.io/docs/concepts/overview/working-with-objects/names/#names
Azure specifies an Azure authentication method for the backend.
Details for managed identity authentication
SecretRef references a Kubernetes Secret containing the Azure credentials. The Secret must have keys "clientId", "tenantId", and "clientSecret".
Name of the referent. This field is effectively required, but due to backwards compatibility is allowed to be empty. Instances of this type with an empty value here are almost certainly wrong. More info: https://kubernetes.io/docs/concepts/overview/working-with-objects/names/#names
Auth specifies to use a Google authentication method for the backend. When omitted, we will try to use the default AWS SDK authentication methods.
audience allows explicitly configuring the `aud` of the ID Token. Ony valid with `IdToken` type. If not set, the aud is automatically derived from the backend hostname.
The type of token to generate. To authenticate to GCP services, generally an AccessToken is used. To authenticate to CloudRun, an IdToken is used.
key provides an inline key to use as the value of the Authorization header. This option is the least secure; usage of a Secret is preferred.
passthrough passes through an existing token that has been sent by the client and validated. Other policies, like JWT and API Key authentication, will strip the original client credentials. Passthrough backend authentication causes the original token to be added back into the request. If there are no client authentication policies on the request, the original token would be unchanged, so this would have no effect.
secretRef references a Kubernetes secret storing the key to use the authorization value. This must be stored in the 'Authorization' key.
Name of the referent. This field is effectively required, but due to backwards compatibility is allowed to be empty. Instances of this type with an empty value here are almost certainly wrong. More info: https://kubernetes.io/docs/concepts/overview/working-with-objects/names/#names
health defines settings for passive and active health checking.
Eviction defines settings for evicting unhealthy backends.
ConsecutiveFailures is the number of consecutive unhealthy responses required before the backend is evicted. For example, a value of 5 means the backend must receive 5 unhealthy responses in a row before being evicted. When both consecutiveFailures and healthThreshold are set, the backend is evicted when either condition is met. When neither is set, a single unhealthy response can trigger eviction.
Duration specifies the base time a backend should be evicted after being marked unhealthy. Subsequent evictions use multiplicative backoff (duration * times_evicted). If all endpoints are evicted, the load balancer falls back to returning evicted endpoints rather than failing entirely. If unset, defaults to 3s.
HealthThreshold is the EWMA (exponentially-weighted moving average) health score threshold, expressed as 0–100. When set, a backend is only evicted if its computed health drops below this value after an unhealthy response. For example, 50 means the backend is evicted when its EWMA health falls below 50% following failures. Unlike consecutiveFailures (which counts consecutive failures), this uses a sliding-window average so a single success in a stream of failures can delay eviction. When both consecutiveFailures and healthThreshold are set, the backend is evicted when either condition is met. When neither is set, a single unhealthy response triggers eviction.
RestoreHealth is the health score (0–100) assigned to a backend when it returns from eviction. For gradual recovery, set below 100; for full recovery immediately, set 100. If unset, the backend resumes with the health it had when evicted.
UnhealthyCondition is a CEL expression that determines whether a response indicates an unhealthy backend. When the expression evaluates to true, the backend is considered unhealthy and may be evicted. For example, to evict on 5xx responses: `response.code >= 500`. When unset, any 5xx response, or a connection failure, is treated as unhealthy. This default lowers the backend's health score but does not trigger eviction on its own.
http defines settings for managing HTTP requests to the backend.
requestTimeout specifies the deadline for receiving a response from the backend.
version specifies the HTTP protocol version to use when connecting to the backend. If not specified, the version is automatically determined: * Service types can specify it with 'appProtocol' on the Service port. * If traffic is identified as gRPC, HTTP2 is used. * If the incoming traffic was plaintext HTTP, the original protocol will be used. * If the incoming traffic was HTTPS, HTTP1 will be used. This is because most clients will transparently upgrade HTTPS traffic to HTTP2, even if the backend doesn't support it
tcp defines settings for managing TCP connections to the backend.
connectTimeout defines the deadline for establishing a connection to the destination.
keepAlive defines settings for enabling TCP keepalives on the connection.
interval specifies the number of seconds between keep-alive probes. If unset, this defaults to 180s.
retries specifies the maximum number of keep-alive probes to send before dropping the connection. If unset, this defaults to 9.
time specifies the number of seconds a connection needs to be idle before keep-alive probes start being sent. If unset, this defaults to 180s.
tls defines settings for managing TLS connections to the backend. If this field is set, TLS will be initiated to the backend; the system trusted CA certificates will be used to validate the server, and the SNI will automatically be set based on the destination.
alpnProtocols sets the Application Level Protocol Negotiation (ALPN) value to use in the TLS handshake. If not present, defaults to ["h2", "http/1.1"].
caCertificateRefs defines the CA certificate ConfigMap to use to verify the server certificate. If unset, the system's trusted certificates are used.
Name of the referent. This field is effectively required, but due to backwards compatibility is allowed to be empty. Instances of this type with an empty value here are almost certainly wrong. More info: https://kubernetes.io/docs/concepts/overview/working-with-objects/names/#names
insecureSkipVerify originates TLS but skips verification of the backend's certificate. WARNING: This is an insecure option that should only be used if the risks are understood. There are two modes: * All disables all TLS verification * Hostname verifies the CA certificate is trusted, but ignores any mismatch of hostname/SANs. Note that this method is still insecure; prefer setting verifySubjectAltNames to customize the valid hostnames if possible.
mtlsCertificateRef enables mutual TLS to the backend, using the specified key (tls.key) and cert (tls.crt) from the refenced Secret. An optional 'ca.cert' field, if present, will be used to verify the server certificate if present. If caCertificateRefs is also specified, the caCertificateRefs field takes priority. If unspecified, no client certificate will be used.
Name of the referent. This field is effectively required, but due to backwards compatibility is allowed to be empty. Instances of this type with an empty value here are almost certainly wrong. More info: https://kubernetes.io/docs/concepts/overview/working-with-objects/names/#names
sni specifies the Server Name Indicator (SNI) to be used in the TLS handshake. If unset, the SNI is automatically set based on the destination hostname.
verifySubjectAltNames specifies the Subject Alternative Names (SAN) to verify in the server certificate. If not present, the destination hostname is automatically used.
transformation is used to mutate and transform requests and responses sent to and from the backend.
request is used to modify the request path.
add is a list of headers to add to the request and what that value should be set to. If there is already a header with these values then append the value as an extra entry.
the name of the header to add.
value is the CEL expression to apply to generate the output value for the header.
body controls manipulation of the HTTP body.
metadata stores CEL-evaluated values under the `metadata` CEL variable for subsequent policy evaluations. metadata is evaluated before header or body transformations.
Remove is a list of header names to remove from the request/response.
set is a list of headers and the value they should be set to.
the name of the header to add.
value is the CEL expression to apply to generate the output value for the header.
response is used to modify the response path.
add is a list of headers to add to the request and what that value should be set to. If there is already a header with these values then append the value as an extra entry.
the name of the header to add.
value is the CEL expression to apply to generate the output value for the header.
body controls manipulation of the HTTP body.
metadata stores CEL-evaluated values under the `metadata` CEL variable for subsequent policy evaluations. metadata is evaluated before header or body transformations.
Remove is a list of header names to remove from the request/response.
set is a list of headers and the value they should be set to.
the name of the header to add.
value is the CEL expression to apply to generate the output value for the header.
tunnel defines settings for managing tunnel connections (like HTTPS_PROXY) to the backend.
backendRef references the proxy server to reach. Supported types: Service and Backend.
Group is the group of the referent. For example, "gateway.networking.k8s.io". When unspecified or empty string, core API group is inferred.
Kind is the Kubernetes resource kind of the referent. For example "Service". Defaults to "Service" when not specified. ExternalName services can refer to CNAME DNS records that may live outside of the cluster and as such are difficult to reason about in terms of conformance. They also may not be safe to forward to (see CVE-2021-25740 for more information). Implementations SHOULD NOT support ExternalName Services. Support: Core (Services with a type other than ExternalName) Support: Implementation-specific (Services with type ExternalName)
Name is the name of the referent.
Namespace is the namespace of the backend. When unspecified, the local namespace is inferred. Note that when a namespace different than the local namespace is specified, a ReferenceGrant object is required in the referent namespace to allow that namespace's owner to accept the reference. See the ReferenceGrant documentation for details. Support: Core
Port specifies the destination port number to use for this resource. Port is required when the referent is a Kubernetes Service. In this case, the port number is the service port number, not the target port. For other resources, destination port might be derived from the referent resource or this field.
Port specifies the port to send the requests to.
Vertex AI provider
Optional: Override the model name, such as `gpt-4o-mini`. If unset, the model name is taken from the request.
The ID of the Google Cloud Project that you use for the Vertex AI.
The location of the Google Cloud Project that you use for the Vertex AI.
provider specifies configuration for how to reach the configured LLM provider.
Anthropic provider
Optional: Override the model name, such as `gpt-4o-mini`. If unset, the model name is taken from the request.
Azure OpenAI provider
The version of the Azure OpenAI API to use. For more information, see the [Azure OpenAI API version reference](https://learn.microsoft.com/en-us/azure/ai-foundry/?view=foundry-classicreference#api-specs). If unset, defaults to "v1"
The name of the Azure OpenAI model deployment to use. For more information, see the [Azure OpenAI model docs](https://learn.microsoft.com/en-us/azure/ai-foundry/foundry-models/concepts/models-sold-directly-by-azure?view=foundry-classic). This is required if ApiVersion is not 'v1'. For v1, the model can be set in the request.
The endpoint for the Azure OpenAI API to use, such as `my-endpoint.openai.azure.com`. If the scheme is included, it is stripped.
Bedrock provider
Guardrail configures the Guardrail policy to use for the backend. See <https://docs.aws.amazon.com/bedrock/latest/userguide/guardrails.html> If not specified, the AWS Guardrail policy will not be used.
GuardrailIdentifier is the identifier of the Guardrail policy to use for the backend.
GuardrailVersion is the version of the Guardrail policy to use for the backend.
Optional: Override the model name, such as `gpt-4o-mini`. If unset, the model name is taken from the request.
Region is the AWS region to use for the backend. Defaults to us-east-1 if not specified.
Gemini provider
Optional: Override the model name, such as `gemini-2.5-pro`. If unset, the model name is taken from the request.
Host specifies the hostname to send the requests to. If not specified, the default hostname for the provider is used.
OpenAI provider
Optional: Override the model name, such as `gpt-4o-mini`. If unset, the model name is taken from the request.
Path specifies the URL path to use for the LLM provider API requests. This is useful when you need to route requests to a different API endpoint while maintaining compatibility with the original provider's API structure. If not specified, the default path for the provider is used.
Port specifies the port to send the requests to.
Vertex AI provider
Optional: Override the model name, such as `gpt-4o-mini`. If unset, the model name is taken from the request.
The ID of the Google Cloud Project that you use for the Vertex AI.
The location of the Google Cloud Project that you use for the Vertex AI.
aws represents an AWS service backend (AgentCore, etc.).
agentCore configures Amazon Bedrock AgentCore as a backend.
agentRuntimeArn is the ARN of the AgentCore runtime.
qualifier optionally specifies the alias or version qualifier.
dynamicForwardProxy configures the proxy to dynamically send requests to the destination based on the incoming request HTTP host header, or TLS SNI for TLS traffic. Note: this Backend type enables users to send trigger the proxy to send requests to arbitrary destinations. Proper access controls must be put in place when using this backend type.
mcp represents an MCP backend
FailureMode controls behavior when MCP targets fail to initialize or become unavailable at runtime. "FailOpen" skips failed targets and continues serving from healthy ones. "FailClosed" (default) fails the entire session if any target fails.
SessionRouting configures MCP session behavior for requests. Defaults to Stateful if not set.
Targets is a list of MCPBackend targets to use for this backend. Policies targeting MCPBackend targets must use targetRefs[].sectionName to select the target by name.
Name of the MCPBackend target.
selector is a label selector is the selector to use to select Services. If policies are needed on a per-service basis, AgentgatewayPolicy can target the desired Service.
namespace is the label selector in which namespaces Services should be selected from. If unset, only the namespace of the AgentgatewayBackend is searched.
matchExpressions is a list of label selector requirements. The requirements are ANDed.
key is the label key that the selector applies to.
operator represents a key's relationship to a set of values. Valid operators are In, NotIn, Exists and DoesNotExist.
values is an array of string values. If the operator is In or NotIn, the values array must be non-empty. If the operator is Exists or DoesNotExist, the values array must be empty. This array is replaced during a strategic merge patch.
matchLabels is a map of {key,value} pairs. A single {key,value} in the matchLabels map is equivalent to an element of matchExpressions, whose key field is "key", the operator is "In", and the values array contains only "value". The requirements are ANDed.
services is the label selector for which Services should be selected.
matchExpressions is a list of label selector requirements. The requirements are ANDed.
key is the label key that the selector applies to.
operator represents a key's relationship to a set of values. Valid operators are In, NotIn, Exists and DoesNotExist.
values is an array of string values. If the operator is In or NotIn, the values array must be non-empty. If the operator is Exists or DoesNotExist, the values array must be empty. This array is replaced during a strategic merge patch.
matchLabels is a map of {key,value} pairs. A single {key,value} in the matchLabels map is equivalent to an element of matchExpressions, whose key field is "key", the operator is "In", and the values array contains only "value". The requirements are ANDed.
static configures a static MCP destination. When connecting to in-cluster Services, it is recommended to use 'selector' instead.
Host is the hostname or IP address of the MCPBackend target.
Path is the URL path of the MCPBackend target endpoint. Defaults to "/sse" for SSE protocol or "/mcp" for StreamableHTTP protocol if not specified.
policies controls policies for communicating with this backend. Policies may also be set in AgentgatewayPolicy, or in the top level AgentgatewayBackend. Policies are merged on a field-level basis, with order: AgentgatewayPolicy < AgentgatewayBackend < AgentgatewayBackend MCP (this field).
auth defines settings for managing authentication to the backend
Auth specifies an explicit AWS authentication method for the backend. When omitted, we will try to use the default AWS SDK authentication methods.
SecretRef references a Kubernetes Secret containing the AWS credentials. The Secret must have keys "accessKey", "secretKey", and optionally "sessionToken".
Name of the referent. This field is effectively required, but due to backwards compatibility is allowed to be empty. Instances of this type with an empty value here are almost certainly wrong. More info: https://kubernetes.io/docs/concepts/overview/working-with-objects/names/#names
Azure specifies an Azure authentication method for the backend.
Details for managed identity authentication
SecretRef references a Kubernetes Secret containing the Azure credentials. The Secret must have keys "clientId", "tenantId", and "clientSecret".
Name of the referent. This field is effectively required, but due to backwards compatibility is allowed to be empty. Instances of this type with an empty value here are almost certainly wrong. More info: https://kubernetes.io/docs/concepts/overview/working-with-objects/names/#names
Auth specifies to use a Google authentication method for the backend. When omitted, we will try to use the default AWS SDK authentication methods.
audience allows explicitly configuring the `aud` of the ID Token. Ony valid with `IdToken` type. If not set, the aud is automatically derived from the backend hostname.
The type of token to generate. To authenticate to GCP services, generally an AccessToken is used. To authenticate to CloudRun, an IdToken is used.
key provides an inline key to use as the value of the Authorization header. This option is the least secure; usage of a Secret is preferred.
passthrough passes through an existing token that has been sent by the client and validated. Other policies, like JWT and API Key authentication, will strip the original client credentials. Passthrough backend authentication causes the original token to be added back into the request. If there are no client authentication policies on the request, the original token would be unchanged, so this would have no effect.
secretRef references a Kubernetes secret storing the key to use the authorization value. This must be stored in the 'Authorization' key.
Name of the referent. This field is effectively required, but due to backwards compatibility is allowed to be empty. Instances of this type with an empty value here are almost certainly wrong. More info: https://kubernetes.io/docs/concepts/overview/working-with-objects/names/#names
health defines settings for passive and active health checking.
Eviction defines settings for evicting unhealthy backends.
ConsecutiveFailures is the number of consecutive unhealthy responses required before the backend is evicted. For example, a value of 5 means the backend must receive 5 unhealthy responses in a row before being evicted. When both consecutiveFailures and healthThreshold are set, the backend is evicted when either condition is met. When neither is set, a single unhealthy response can trigger eviction.
Duration specifies the base time a backend should be evicted after being marked unhealthy. Subsequent evictions use multiplicative backoff (duration * times_evicted). If all endpoints are evicted, the load balancer falls back to returning evicted endpoints rather than failing entirely. If unset, defaults to 3s.
HealthThreshold is the EWMA (exponentially-weighted moving average) health score threshold, expressed as 0–100. When set, a backend is only evicted if its computed health drops below this value after an unhealthy response. For example, 50 means the backend is evicted when its EWMA health falls below 50% following failures. Unlike consecutiveFailures (which counts consecutive failures), this uses a sliding-window average so a single success in a stream of failures can delay eviction. When both consecutiveFailures and healthThreshold are set, the backend is evicted when either condition is met. When neither is set, a single unhealthy response triggers eviction.
RestoreHealth is the health score (0–100) assigned to a backend when it returns from eviction. For gradual recovery, set below 100; for full recovery immediately, set 100. If unset, the backend resumes with the health it had when evicted.
UnhealthyCondition is a CEL expression that determines whether a response indicates an unhealthy backend. When the expression evaluates to true, the backend is considered unhealthy and may be evicted. For example, to evict on 5xx responses: `response.code >= 500`. When unset, any 5xx response, or a connection failure, is treated as unhealthy. This default lowers the backend's health score but does not trigger eviction on its own.
http defines settings for managing HTTP requests to the backend.
requestTimeout specifies the deadline for receiving a response from the backend.
version specifies the HTTP protocol version to use when connecting to the backend. If not specified, the version is automatically determined: * Service types can specify it with 'appProtocol' on the Service port. * If traffic is identified as gRPC, HTTP2 is used. * If the incoming traffic was plaintext HTTP, the original protocol will be used. * If the incoming traffic was HTTPS, HTTP1 will be used. This is because most clients will transparently upgrade HTTPS traffic to HTTP2, even if the backend doesn't support it
mcp specifies settings for MCP workloads. This is only applicable when connecting to a Backend of type 'mcp'.
authentication defines MCPBackend specific authentication rules.
audiences specify the list of allowed audiences that are allowed access. This corresponds to the 'aud' claim (https://datatracker.ietf.org/doc/html/rfc7519#section-4.1.3). If unset, any audience is allowed.
Issuer identifies the IdP that issued the JWT. This corresponds to the 'iss' claim (https://tools.ietf.org/html/rfc7519#section-4.1.1).
jwks defines the remote JSON Web Key used to validate the signature of the JWT.
backendRef references the remote JWKS server to reach. Supported types are Service and (static) Backend. An AgentgatewayPolicy containing backend tls config can then be attached to the service/backend in order to set tls options for a connection to the remote jwks source.
Group is the group of the referent. For example, "gateway.networking.k8s.io". When unspecified or empty string, core API group is inferred.
Kind is the Kubernetes resource kind of the referent. For example "Service". Defaults to "Service" when not specified. ExternalName services can refer to CNAME DNS records that may live outside of the cluster and as such are difficult to reason about in terms of conformance. They also may not be safe to forward to (see CVE-2021-25740 for more information). Implementations SHOULD NOT support ExternalName Services. Support: Core (Services with a type other than ExternalName) Support: Implementation-specific (Services with type ExternalName)
Name is the name of the referent.
Namespace is the namespace of the backend. When unspecified, the local namespace is inferred. Note that when a namespace different than the local namespace is specified, a ReferenceGrant object is required in the referent namespace to allow that namespace's owner to accept the reference. See the ReferenceGrant documentation for details. Support: Core
Port specifies the destination port number to use for this resource. Port is required when the referent is a Kubernetes Service. In this case, the port number is the service port number, not the target port. For other resources, destination port might be derived from the referent resource or this field.
Path to IdP jwks endpoint, relative to the root, commonly ".well-known/jwks.json".
validation mode for JWT authentication.
McpIDP specifies the identity provider to use for authentication
ResourceMetadata defines the metadata to use for MCP resources.
authorization defines MCPBackend level authorization. Unlike authorization at the HTTP level, which will reject unauthorized requests with a 403 error, this policy works at the MCPBackend level. List operations, such as list_tools, will have each item evaluated. Items that do not meet the rule will be filtered. Get or call operations, such as call_tool, will evaluate the specific item and reject requests that do not meet the rule.
Action defines whether the rule allows or denies the request if matched. If unspecified, the default is "Allow".
Policy specifies the Authorization rule to evaluate. A policy matches when **any** of the conditions evaluates to true.
MatchExpressions defines a set of conditions that must be satisfied for the rule to match. These expression should be in the form of a Common Expression Language (CEL) expression.
tcp defines settings for managing TCP connections to the backend.
connectTimeout defines the deadline for establishing a connection to the destination.
keepAlive defines settings for enabling TCP keepalives on the connection.
interval specifies the number of seconds between keep-alive probes. If unset, this defaults to 180s.
retries specifies the maximum number of keep-alive probes to send before dropping the connection. If unset, this defaults to 9.
time specifies the number of seconds a connection needs to be idle before keep-alive probes start being sent. If unset, this defaults to 180s.
tls defines settings for managing TLS connections to the backend. If this field is set, TLS will be initiated to the backend; the system trusted CA certificates will be used to validate the server, and the SNI will automatically be set based on the destination.
alpnProtocols sets the Application Level Protocol Negotiation (ALPN) value to use in the TLS handshake. If not present, defaults to ["h2", "http/1.1"].
caCertificateRefs defines the CA certificate ConfigMap to use to verify the server certificate. If unset, the system's trusted certificates are used.
Name of the referent. This field is effectively required, but due to backwards compatibility is allowed to be empty. Instances of this type with an empty value here are almost certainly wrong. More info: https://kubernetes.io/docs/concepts/overview/working-with-objects/names/#names
insecureSkipVerify originates TLS but skips verification of the backend's certificate. WARNING: This is an insecure option that should only be used if the risks are understood. There are two modes: * All disables all TLS verification * Hostname verifies the CA certificate is trusted, but ignores any mismatch of hostname/SANs. Note that this method is still insecure; prefer setting verifySubjectAltNames to customize the valid hostnames if possible.
mtlsCertificateRef enables mutual TLS to the backend, using the specified key (tls.key) and cert (tls.crt) from the refenced Secret. An optional 'ca.cert' field, if present, will be used to verify the server certificate if present. If caCertificateRefs is also specified, the caCertificateRefs field takes priority. If unspecified, no client certificate will be used.
Name of the referent. This field is effectively required, but due to backwards compatibility is allowed to be empty. Instances of this type with an empty value here are almost certainly wrong. More info: https://kubernetes.io/docs/concepts/overview/working-with-objects/names/#names
sni specifies the Server Name Indicator (SNI) to be used in the TLS handshake. If unset, the SNI is automatically set based on the destination hostname.
verifySubjectAltNames specifies the Subject Alternative Names (SAN) to verify in the server certificate. If not present, the destination hostname is automatically used.
transformation is used to mutate and transform requests and responses sent to and from the backend.
request is used to modify the request path.
add is a list of headers to add to the request and what that value should be set to. If there is already a header with these values then append the value as an extra entry.
the name of the header to add.
value is the CEL expression to apply to generate the output value for the header.
body controls manipulation of the HTTP body.
metadata stores CEL-evaluated values under the `metadata` CEL variable for subsequent policy evaluations. metadata is evaluated before header or body transformations.
Remove is a list of header names to remove from the request/response.
set is a list of headers and the value they should be set to.
the name of the header to add.
value is the CEL expression to apply to generate the output value for the header.
response is used to modify the response path.
add is a list of headers to add to the request and what that value should be set to. If there is already a header with these values then append the value as an extra entry.
the name of the header to add.
value is the CEL expression to apply to generate the output value for the header.
body controls manipulation of the HTTP body.
metadata stores CEL-evaluated values under the `metadata` CEL variable for subsequent policy evaluations. metadata is evaluated before header or body transformations.
Remove is a list of header names to remove from the request/response.
set is a list of headers and the value they should be set to.
the name of the header to add.
value is the CEL expression to apply to generate the output value for the header.
tunnel defines settings for managing tunnel connections (like HTTPS_PROXY) to the backend.
backendRef references the proxy server to reach. Supported types: Service and Backend.
Group is the group of the referent. For example, "gateway.networking.k8s.io". When unspecified or empty string, core API group is inferred.
Kind is the Kubernetes resource kind of the referent. For example "Service". Defaults to "Service" when not specified. ExternalName services can refer to CNAME DNS records that may live outside of the cluster and as such are difficult to reason about in terms of conformance. They also may not be safe to forward to (see CVE-2021-25740 for more information). Implementations SHOULD NOT support ExternalName Services. Support: Core (Services with a type other than ExternalName) Support: Implementation-specific (Services with type ExternalName)
Name is the name of the referent.
Namespace is the namespace of the backend. When unspecified, the local namespace is inferred. Note that when a namespace different than the local namespace is specified, a ReferenceGrant object is required in the referent namespace to allow that namespace's owner to accept the reference. See the ReferenceGrant documentation for details. Support: Core
Port specifies the destination port number to use for this resource. Port is required when the referent is a Kubernetes Service. In this case, the port number is the service port number, not the target port. For other resources, destination port might be derived from the referent resource or this field.
Port is the port number of the MCPBackend target.
Protocol is the protocol to use for the connection to the MCPBackend target.
policies controls policies for communicating with this backend. Policies may also be set in AgentgatewayPolicy; policies are merged on a field-level basis, with policies on the Backend (this field) taking precedence.
ai specifies settings for AI workloads. This is only applicable when connecting to a Backend of type 'ai'.
Provide defaults to merge with user input fields. If the field is already set, the field in the request is used.
The name of the field.
The field default value, which can be any JSON Data Type.
ModelAliases maps friendly model names to actual provider model names. Example: {"fast": "gpt-3.5-turbo", "smart": "gpt-4-turbo"} Note: This field is only applicable when using the agentgateway data plane.
Provide overrides to merge with user input fields. If the field is already set, the field will be overwritten.
The name of the field.
The field default value, which can be any JSON Data Type.
Enrich requests sent to the LLM provider by appending and prepending system prompts. This can be configured only for LLM providers that use the `CHAT` or `CHAT_STREAMING` API route type.
A list of messages to be appended to the prompt sent by the client.
String content of the message.
Role of the message. The available roles depend on the backend LLM provider model, such as `SYSTEM` or `USER` in the OpenAI API.
A list of messages to be prepended to the prompt sent by the client.
String content of the message.
Role of the message. The available roles depend on the backend LLM provider model, such as `SYSTEM` or `USER` in the OpenAI API.
promptCaching enables automatic prompt caching for supported providers (AWS Bedrock). Reduces API costs by caching static content like system prompts and tool definitions. Only applicable for Bedrock Claude 3+ and Nova models.
CacheMessages enables caching for conversation messages. Caches all messages in the conversation for cost savings.
CacheSystem enables caching for system prompts. Inserts a cache point after all system messages.
CacheTools enables caching for tool definitions. Inserts a cache point after all tool specifications.
MinTokens specifies the minimum estimated token count before caching is enabled. Uses rough heuristic (word count × 1.3) to estimate tokens. Bedrock requires at least 1,024 tokens for caching to be effective.
promptGuard enables adding guardrails to LLM requests and responses.
Prompt guards to apply to requests sent by the client.
bedrockGuardrails configures AWS Bedrock Guardrails for prompt guarding.
GuardrailIdentifier is the identifier of the Guardrail policy to use for the backend.
policies controls policies for communicating with AWS Bedrock Guardrails.
auth defines settings for managing authentication to the backend
Auth specifies an explicit AWS authentication method for the backend. When omitted, we will try to use the default AWS SDK authentication methods.
SecretRef references a Kubernetes Secret containing the AWS credentials. The Secret must have keys "accessKey", "secretKey", and optionally "sessionToken".
Name of the referent. This field is effectively required, but due to backwards compatibility is allowed to be empty. Instances of this type with an empty value here are almost certainly wrong. More info: https://kubernetes.io/docs/concepts/overview/working-with-objects/names/#names
Azure specifies an Azure authentication method for the backend.
Details for managed identity authentication
SecretRef references a Kubernetes Secret containing the Azure credentials. The Secret must have keys "clientId", "tenantId", and "clientSecret".
Name of the referent. This field is effectively required, but due to backwards compatibility is allowed to be empty. Instances of this type with an empty value here are almost certainly wrong. More info: https://kubernetes.io/docs/concepts/overview/working-with-objects/names/#names
Auth specifies to use a Google authentication method for the backend. When omitted, we will try to use the default AWS SDK authentication methods.
audience allows explicitly configuring the `aud` of the ID Token. Ony valid with `IdToken` type. If not set, the aud is automatically derived from the backend hostname.
The type of token to generate. To authenticate to GCP services, generally an AccessToken is used. To authenticate to CloudRun, an IdToken is used.
key provides an inline key to use as the value of the Authorization header. This option is the least secure; usage of a Secret is preferred.
passthrough passes through an existing token that has been sent by the client and validated. Other policies, like JWT and API Key authentication, will strip the original client credentials. Passthrough backend authentication causes the original token to be added back into the request. If there are no client authentication policies on the request, the original token would be unchanged, so this would have no effect.
secretRef references a Kubernetes secret storing the key to use the authorization value. This must be stored in the 'Authorization' key.
Name of the referent. This field is effectively required, but due to backwards compatibility is allowed to be empty. Instances of this type with an empty value here are almost certainly wrong. More info: https://kubernetes.io/docs/concepts/overview/working-with-objects/names/#names
health defines settings for passive and active health checking.
Eviction defines settings for evicting unhealthy backends.
ConsecutiveFailures is the number of consecutive unhealthy responses required before the backend is evicted. For example, a value of 5 means the backend must receive 5 unhealthy responses in a row before being evicted. When both consecutiveFailures and healthThreshold are set, the backend is evicted when either condition is met. When neither is set, a single unhealthy response can trigger eviction.
Duration specifies the base time a backend should be evicted after being marked unhealthy. Subsequent evictions use multiplicative backoff (duration * times_evicted). If all endpoints are evicted, the load balancer falls back to returning evicted endpoints rather than failing entirely. If unset, defaults to 3s.
HealthThreshold is the EWMA (exponentially-weighted moving average) health score threshold, expressed as 0–100. When set, a backend is only evicted if its computed health drops below this value after an unhealthy response. For example, 50 means the backend is evicted when its EWMA health falls below 50% following failures. Unlike consecutiveFailures (which counts consecutive failures), this uses a sliding-window average so a single success in a stream of failures can delay eviction. When both consecutiveFailures and healthThreshold are set, the backend is evicted when either condition is met. When neither is set, a single unhealthy response triggers eviction.
RestoreHealth is the health score (0–100) assigned to a backend when it returns from eviction. For gradual recovery, set below 100; for full recovery immediately, set 100. If unset, the backend resumes with the health it had when evicted.
UnhealthyCondition is a CEL expression that determines whether a response indicates an unhealthy backend. When the expression evaluates to true, the backend is considered unhealthy and may be evicted. For example, to evict on 5xx responses: `response.code >= 500`. When unset, any 5xx response, or a connection failure, is treated as unhealthy. This default lowers the backend's health score but does not trigger eviction on its own.
http defines settings for managing HTTP requests to the backend.
requestTimeout specifies the deadline for receiving a response from the backend.
version specifies the HTTP protocol version to use when connecting to the backend. If not specified, the version is automatically determined: * Service types can specify it with 'appProtocol' on the Service port. * If traffic is identified as gRPC, HTTP2 is used. * If the incoming traffic was plaintext HTTP, the original protocol will be used. * If the incoming traffic was HTTPS, HTTP1 will be used. This is because most clients will transparently upgrade HTTPS traffic to HTTP2, even if the backend doesn't support it
tcp defines settings for managing TCP connections to the backend.
connectTimeout defines the deadline for establishing a connection to the destination.
keepAlive defines settings for enabling TCP keepalives on the connection.
interval specifies the number of seconds between keep-alive probes. If unset, this defaults to 180s.
retries specifies the maximum number of keep-alive probes to send before dropping the connection. If unset, this defaults to 9.
time specifies the number of seconds a connection needs to be idle before keep-alive probes start being sent. If unset, this defaults to 180s.
tls defines settings for managing TLS connections to the backend. If this field is set, TLS will be initiated to the backend; the system trusted CA certificates will be used to validate the server, and the SNI will automatically be set based on the destination.
alpnProtocols sets the Application Level Protocol Negotiation (ALPN) value to use in the TLS handshake. If not present, defaults to ["h2", "http/1.1"].
caCertificateRefs defines the CA certificate ConfigMap to use to verify the server certificate. If unset, the system's trusted certificates are used.
Name of the referent. This field is effectively required, but due to backwards compatibility is allowed to be empty. Instances of this type with an empty value here are almost certainly wrong. More info: https://kubernetes.io/docs/concepts/overview/working-with-objects/names/#names
insecureSkipVerify originates TLS but skips verification of the backend's certificate. WARNING: This is an insecure option that should only be used if the risks are understood. There are two modes: * All disables all TLS verification * Hostname verifies the CA certificate is trusted, but ignores any mismatch of hostname/SANs. Note that this method is still insecure; prefer setting verifySubjectAltNames to customize the valid hostnames if possible.
mtlsCertificateRef enables mutual TLS to the backend, using the specified key (tls.key) and cert (tls.crt) from the refenced Secret. An optional 'ca.cert' field, if present, will be used to verify the server certificate if present. If caCertificateRefs is also specified, the caCertificateRefs field takes priority. If unspecified, no client certificate will be used.
Name of the referent. This field is effectively required, but due to backwards compatibility is allowed to be empty. Instances of this type with an empty value here are almost certainly wrong. More info: https://kubernetes.io/docs/concepts/overview/working-with-objects/names/#names
sni specifies the Server Name Indicator (SNI) to be used in the TLS handshake. If unset, the SNI is automatically set based on the destination hostname.
verifySubjectAltNames specifies the Subject Alternative Names (SAN) to verify in the server certificate. If not present, the destination hostname is automatically used.
transformation is used to mutate and transform requests and responses sent to and from the backend.
request is used to modify the request path.
add is a list of headers to add to the request and what that value should be set to. If there is already a header with these values then append the value as an extra entry.
the name of the header to add.
value is the CEL expression to apply to generate the output value for the header.
body controls manipulation of the HTTP body.
metadata stores CEL-evaluated values under the `metadata` CEL variable for subsequent policy evaluations. metadata is evaluated before header or body transformations.
Remove is a list of header names to remove from the request/response.
set is a list of headers and the value they should be set to.
the name of the header to add.
value is the CEL expression to apply to generate the output value for the header.
response is used to modify the response path.
add is a list of headers to add to the request and what that value should be set to. If there is already a header with these values then append the value as an extra entry.
the name of the header to add.
value is the CEL expression to apply to generate the output value for the header.
body controls manipulation of the HTTP body.
metadata stores CEL-evaluated values under the `metadata` CEL variable for subsequent policy evaluations. metadata is evaluated before header or body transformations.
Remove is a list of header names to remove from the request/response.
set is a list of headers and the value they should be set to.
the name of the header to add.
value is the CEL expression to apply to generate the output value for the header.
tunnel defines settings for managing tunnel connections (like HTTPS_PROXY) to the backend.
backendRef references the proxy server to reach. Supported types: Service and Backend.
Group is the group of the referent. For example, "gateway.networking.k8s.io". When unspecified or empty string, core API group is inferred.
Kind is the Kubernetes resource kind of the referent. For example "Service". Defaults to "Service" when not specified. ExternalName services can refer to CNAME DNS records that may live outside of the cluster and as such are difficult to reason about in terms of conformance. They also may not be safe to forward to (see CVE-2021-25740 for more information). Implementations SHOULD NOT support ExternalName Services. Support: Core (Services with a type other than ExternalName) Support: Implementation-specific (Services with type ExternalName)
Name is the name of the referent.
Namespace is the namespace of the backend. When unspecified, the local namespace is inferred. Note that when a namespace different than the local namespace is specified, a ReferenceGrant object is required in the referent namespace to allow that namespace's owner to accept the reference. See the ReferenceGrant documentation for details. Support: Core
Port specifies the destination port number to use for this resource. Port is required when the referent is a Kubernetes Service. In this case, the port number is the service port number, not the target port. For other resources, destination port might be derived from the referent resource or this field.
Region is the AWS region where the guardrail is deployed (e.g., "us-west-2").
GuardrailVersion is the version of the Guardrail policy to use for the backend.
googleModelArmor configures Google Model Armor for prompt guarding.
Location is the Google Cloud location (e.g., "us-central1"). Defaults to "us-central1" if not specified.
policies controls policies for communicating with Google Model Armor.
auth defines settings for managing authentication to the backend
Auth specifies an explicit AWS authentication method for the backend. When omitted, we will try to use the default AWS SDK authentication methods.
SecretRef references a Kubernetes Secret containing the AWS credentials. The Secret must have keys "accessKey", "secretKey", and optionally "sessionToken".
Name of the referent. This field is effectively required, but due to backwards compatibility is allowed to be empty. Instances of this type with an empty value here are almost certainly wrong. More info: https://kubernetes.io/docs/concepts/overview/working-with-objects/names/#names
Azure specifies an Azure authentication method for the backend.
Details for managed identity authentication
SecretRef references a Kubernetes Secret containing the Azure credentials. The Secret must have keys "clientId", "tenantId", and "clientSecret".
Name of the referent. This field is effectively required, but due to backwards compatibility is allowed to be empty. Instances of this type with an empty value here are almost certainly wrong. More info: https://kubernetes.io/docs/concepts/overview/working-with-objects/names/#names
Auth specifies to use a Google authentication method for the backend. When omitted, we will try to use the default AWS SDK authentication methods.
audience allows explicitly configuring the `aud` of the ID Token. Ony valid with `IdToken` type. If not set, the aud is automatically derived from the backend hostname.
The type of token to generate. To authenticate to GCP services, generally an AccessToken is used. To authenticate to CloudRun, an IdToken is used.
key provides an inline key to use as the value of the Authorization header. This option is the least secure; usage of a Secret is preferred.
passthrough passes through an existing token that has been sent by the client and validated. Other policies, like JWT and API Key authentication, will strip the original client credentials. Passthrough backend authentication causes the original token to be added back into the request. If there are no client authentication policies on the request, the original token would be unchanged, so this would have no effect.
secretRef references a Kubernetes secret storing the key to use the authorization value. This must be stored in the 'Authorization' key.
Name of the referent. This field is effectively required, but due to backwards compatibility is allowed to be empty. Instances of this type with an empty value here are almost certainly wrong. More info: https://kubernetes.io/docs/concepts/overview/working-with-objects/names/#names
health defines settings for passive and active health checking.
Eviction defines settings for evicting unhealthy backends.
ConsecutiveFailures is the number of consecutive unhealthy responses required before the backend is evicted. For example, a value of 5 means the backend must receive 5 unhealthy responses in a row before being evicted. When both consecutiveFailures and healthThreshold are set, the backend is evicted when either condition is met. When neither is set, a single unhealthy response can trigger eviction.
Duration specifies the base time a backend should be evicted after being marked unhealthy. Subsequent evictions use multiplicative backoff (duration * times_evicted). If all endpoints are evicted, the load balancer falls back to returning evicted endpoints rather than failing entirely. If unset, defaults to 3s.
HealthThreshold is the EWMA (exponentially-weighted moving average) health score threshold, expressed as 0–100. When set, a backend is only evicted if its computed health drops below this value after an unhealthy response. For example, 50 means the backend is evicted when its EWMA health falls below 50% following failures. Unlike consecutiveFailures (which counts consecutive failures), this uses a sliding-window average so a single success in a stream of failures can delay eviction. When both consecutiveFailures and healthThreshold are set, the backend is evicted when either condition is met. When neither is set, a single unhealthy response triggers eviction.
RestoreHealth is the health score (0–100) assigned to a backend when it returns from eviction. For gradual recovery, set below 100; for full recovery immediately, set 100. If unset, the backend resumes with the health it had when evicted.
UnhealthyCondition is a CEL expression that determines whether a response indicates an unhealthy backend. When the expression evaluates to true, the backend is considered unhealthy and may be evicted. For example, to evict on 5xx responses: `response.code >= 500`. When unset, any 5xx response, or a connection failure, is treated as unhealthy. This default lowers the backend's health score but does not trigger eviction on its own.
http defines settings for managing HTTP requests to the backend.
requestTimeout specifies the deadline for receiving a response from the backend.
version specifies the HTTP protocol version to use when connecting to the backend. If not specified, the version is automatically determined: * Service types can specify it with 'appProtocol' on the Service port. * If traffic is identified as gRPC, HTTP2 is used. * If the incoming traffic was plaintext HTTP, the original protocol will be used. * If the incoming traffic was HTTPS, HTTP1 will be used. This is because most clients will transparently upgrade HTTPS traffic to HTTP2, even if the backend doesn't support it
tcp defines settings for managing TCP connections to the backend.
connectTimeout defines the deadline for establishing a connection to the destination.
keepAlive defines settings for enabling TCP keepalives on the connection.
interval specifies the number of seconds between keep-alive probes. If unset, this defaults to 180s.
retries specifies the maximum number of keep-alive probes to send before dropping the connection. If unset, this defaults to 9.
time specifies the number of seconds a connection needs to be idle before keep-alive probes start being sent. If unset, this defaults to 180s.
tls defines settings for managing TLS connections to the backend. If this field is set, TLS will be initiated to the backend; the system trusted CA certificates will be used to validate the server, and the SNI will automatically be set based on the destination.
alpnProtocols sets the Application Level Protocol Negotiation (ALPN) value to use in the TLS handshake. If not present, defaults to ["h2", "http/1.1"].
caCertificateRefs defines the CA certificate ConfigMap to use to verify the server certificate. If unset, the system's trusted certificates are used.
Name of the referent. This field is effectively required, but due to backwards compatibility is allowed to be empty. Instances of this type with an empty value here are almost certainly wrong. More info: https://kubernetes.io/docs/concepts/overview/working-with-objects/names/#names
insecureSkipVerify originates TLS but skips verification of the backend's certificate. WARNING: This is an insecure option that should only be used if the risks are understood. There are two modes: * All disables all TLS verification * Hostname verifies the CA certificate is trusted, but ignores any mismatch of hostname/SANs. Note that this method is still insecure; prefer setting verifySubjectAltNames to customize the valid hostnames if possible.
mtlsCertificateRef enables mutual TLS to the backend, using the specified key (tls.key) and cert (tls.crt) from the refenced Secret. An optional 'ca.cert' field, if present, will be used to verify the server certificate if present. If caCertificateRefs is also specified, the caCertificateRefs field takes priority. If unspecified, no client certificate will be used.
Name of the referent. This field is effectively required, but due to backwards compatibility is allowed to be empty. Instances of this type with an empty value here are almost certainly wrong. More info: https://kubernetes.io/docs/concepts/overview/working-with-objects/names/#names
sni specifies the Server Name Indicator (SNI) to be used in the TLS handshake. If unset, the SNI is automatically set based on the destination hostname.
verifySubjectAltNames specifies the Subject Alternative Names (SAN) to verify in the server certificate. If not present, the destination hostname is automatically used.
transformation is used to mutate and transform requests and responses sent to and from the backend.
request is used to modify the request path.
add is a list of headers to add to the request and what that value should be set to. If there is already a header with these values then append the value as an extra entry.
the name of the header to add.
value is the CEL expression to apply to generate the output value for the header.
body controls manipulation of the HTTP body.
metadata stores CEL-evaluated values under the `metadata` CEL variable for subsequent policy evaluations. metadata is evaluated before header or body transformations.
Remove is a list of header names to remove from the request/response.
set is a list of headers and the value they should be set to.
the name of the header to add.
value is the CEL expression to apply to generate the output value for the header.
response is used to modify the response path.
add is a list of headers to add to the request and what that value should be set to. If there is already a header with these values then append the value as an extra entry.
the name of the header to add.
value is the CEL expression to apply to generate the output value for the header.
body controls manipulation of the HTTP body.
metadata stores CEL-evaluated values under the `metadata` CEL variable for subsequent policy evaluations. metadata is evaluated before header or body transformations.
Remove is a list of header names to remove from the request/response.
set is a list of headers and the value they should be set to.
the name of the header to add.
value is the CEL expression to apply to generate the output value for the header.
tunnel defines settings for managing tunnel connections (like HTTPS_PROXY) to the backend.
backendRef references the proxy server to reach. Supported types: Service and Backend.
Group is the group of the referent. For example, "gateway.networking.k8s.io". When unspecified or empty string, core API group is inferred.
Kind is the Kubernetes resource kind of the referent. For example "Service". Defaults to "Service" when not specified. ExternalName services can refer to CNAME DNS records that may live outside of the cluster and as such are difficult to reason about in terms of conformance. They also may not be safe to forward to (see CVE-2021-25740 for more information). Implementations SHOULD NOT support ExternalName Services. Support: Core (Services with a type other than ExternalName) Support: Implementation-specific (Services with type ExternalName)
Name is the name of the referent.
Namespace is the namespace of the backend. When unspecified, the local namespace is inferred. Note that when a namespace different than the local namespace is specified, a ReferenceGrant object is required in the referent namespace to allow that namespace's owner to accept the reference. See the ReferenceGrant documentation for details. Support: Core
Port specifies the destination port number to use for this resource. Port is required when the referent is a Kubernetes Service. In this case, the port number is the service port number, not the target port. For other resources, destination port might be derived from the referent resource or this field.
ProjectID is the Google Cloud project ID.
TemplateID is the template ID for Google Model Armor.
openAIModeration passes prompt data through the OpenAI Moderations endpoint. See https://developers.openai.com/api/reference/resources/moderations for more information.
model specifies the moderation model to use. For example, `omni-moderation`.
policies controls policies for communicating with OpenAI.
auth defines settings for managing authentication to the backend
Auth specifies an explicit AWS authentication method for the backend. When omitted, we will try to use the default AWS SDK authentication methods.
SecretRef references a Kubernetes Secret containing the AWS credentials. The Secret must have keys "accessKey", "secretKey", and optionally "sessionToken".
Name of the referent. This field is effectively required, but due to backwards compatibility is allowed to be empty. Instances of this type with an empty value here are almost certainly wrong. More info: https://kubernetes.io/docs/concepts/overview/working-with-objects/names/#names
Azure specifies an Azure authentication method for the backend.
Details for managed identity authentication
SecretRef references a Kubernetes Secret containing the Azure credentials. The Secret must have keys "clientId", "tenantId", and "clientSecret".
Name of the referent. This field is effectively required, but due to backwards compatibility is allowed to be empty. Instances of this type with an empty value here are almost certainly wrong. More info: https://kubernetes.io/docs/concepts/overview/working-with-objects/names/#names
Auth specifies to use a Google authentication method for the backend. When omitted, we will try to use the default AWS SDK authentication methods.
audience allows explicitly configuring the `aud` of the ID Token. Ony valid with `IdToken` type. If not set, the aud is automatically derived from the backend hostname.
The type of token to generate. To authenticate to GCP services, generally an AccessToken is used. To authenticate to CloudRun, an IdToken is used.
key provides an inline key to use as the value of the Authorization header. This option is the least secure; usage of a Secret is preferred.
passthrough passes through an existing token that has been sent by the client and validated. Other policies, like JWT and API Key authentication, will strip the original client credentials. Passthrough backend authentication causes the original token to be added back into the request. If there are no client authentication policies on the request, the original token would be unchanged, so this would have no effect.
secretRef references a Kubernetes secret storing the key to use the authorization value. This must be stored in the 'Authorization' key.
Name of the referent. This field is effectively required, but due to backwards compatibility is allowed to be empty. Instances of this type with an empty value here are almost certainly wrong. More info: https://kubernetes.io/docs/concepts/overview/working-with-objects/names/#names
health defines settings for passive and active health checking.
Eviction defines settings for evicting unhealthy backends.
ConsecutiveFailures is the number of consecutive unhealthy responses required before the backend is evicted. For example, a value of 5 means the backend must receive 5 unhealthy responses in a row before being evicted. When both consecutiveFailures and healthThreshold are set, the backend is evicted when either condition is met. When neither is set, a single unhealthy response can trigger eviction.
Duration specifies the base time a backend should be evicted after being marked unhealthy. Subsequent evictions use multiplicative backoff (duration * times_evicted). If all endpoints are evicted, the load balancer falls back to returning evicted endpoints rather than failing entirely. If unset, defaults to 3s.
HealthThreshold is the EWMA (exponentially-weighted moving average) health score threshold, expressed as 0–100. When set, a backend is only evicted if its computed health drops below this value after an unhealthy response. For example, 50 means the backend is evicted when its EWMA health falls below 50% following failures. Unlike consecutiveFailures (which counts consecutive failures), this uses a sliding-window average so a single success in a stream of failures can delay eviction. When both consecutiveFailures and healthThreshold are set, the backend is evicted when either condition is met. When neither is set, a single unhealthy response triggers eviction.
RestoreHealth is the health score (0–100) assigned to a backend when it returns from eviction. For gradual recovery, set below 100; for full recovery immediately, set 100. If unset, the backend resumes with the health it had when evicted.
UnhealthyCondition is a CEL expression that determines whether a response indicates an unhealthy backend. When the expression evaluates to true, the backend is considered unhealthy and may be evicted. For example, to evict on 5xx responses: `response.code >= 500`. When unset, any 5xx response, or a connection failure, is treated as unhealthy. This default lowers the backend's health score but does not trigger eviction on its own.
http defines settings for managing HTTP requests to the backend.
requestTimeout specifies the deadline for receiving a response from the backend.
version specifies the HTTP protocol version to use when connecting to the backend. If not specified, the version is automatically determined: * Service types can specify it with 'appProtocol' on the Service port. * If traffic is identified as gRPC, HTTP2 is used. * If the incoming traffic was plaintext HTTP, the original protocol will be used. * If the incoming traffic was HTTPS, HTTP1 will be used. This is because most clients will transparently upgrade HTTPS traffic to HTTP2, even if the backend doesn't support it
tcp defines settings for managing TCP connections to the backend.
connectTimeout defines the deadline for establishing a connection to the destination.
keepAlive defines settings for enabling TCP keepalives on the connection.
interval specifies the number of seconds between keep-alive probes. If unset, this defaults to 180s.
retries specifies the maximum number of keep-alive probes to send before dropping the connection. If unset, this defaults to 9.
time specifies the number of seconds a connection needs to be idle before keep-alive probes start being sent. If unset, this defaults to 180s.
tls defines settings for managing TLS connections to the backend. If this field is set, TLS will be initiated to the backend; the system trusted CA certificates will be used to validate the server, and the SNI will automatically be set based on the destination.
alpnProtocols sets the Application Level Protocol Negotiation (ALPN) value to use in the TLS handshake. If not present, defaults to ["h2", "http/1.1"].
caCertificateRefs defines the CA certificate ConfigMap to use to verify the server certificate. If unset, the system's trusted certificates are used.
Name of the referent. This field is effectively required, but due to backwards compatibility is allowed to be empty. Instances of this type with an empty value here are almost certainly wrong. More info: https://kubernetes.io/docs/concepts/overview/working-with-objects/names/#names
insecureSkipVerify originates TLS but skips verification of the backend's certificate. WARNING: This is an insecure option that should only be used if the risks are understood. There are two modes: * All disables all TLS verification * Hostname verifies the CA certificate is trusted, but ignores any mismatch of hostname/SANs. Note that this method is still insecure; prefer setting verifySubjectAltNames to customize the valid hostnames if possible.
mtlsCertificateRef enables mutual TLS to the backend, using the specified key (tls.key) and cert (tls.crt) from the refenced Secret. An optional 'ca.cert' field, if present, will be used to verify the server certificate if present. If caCertificateRefs is also specified, the caCertificateRefs field takes priority. If unspecified, no client certificate will be used.
Name of the referent. This field is effectively required, but due to backwards compatibility is allowed to be empty. Instances of this type with an empty value here are almost certainly wrong. More info: https://kubernetes.io/docs/concepts/overview/working-with-objects/names/#names
sni specifies the Server Name Indicator (SNI) to be used in the TLS handshake. If unset, the SNI is automatically set based on the destination hostname.
verifySubjectAltNames specifies the Subject Alternative Names (SAN) to verify in the server certificate. If not present, the destination hostname is automatically used.
transformation is used to mutate and transform requests and responses sent to and from the backend.
request is used to modify the request path.
add is a list of headers to add to the request and what that value should be set to. If there is already a header with these values then append the value as an extra entry.
the name of the header to add.
value is the CEL expression to apply to generate the output value for the header.
body controls manipulation of the HTTP body.
metadata stores CEL-evaluated values under the `metadata` CEL variable for subsequent policy evaluations. metadata is evaluated before header or body transformations.
Remove is a list of header names to remove from the request/response.
set is a list of headers and the value they should be set to.
the name of the header to add.
value is the CEL expression to apply to generate the output value for the header.
response is used to modify the response path.
add is a list of headers to add to the request and what that value should be set to. If there is already a header with these values then append the value as an extra entry.
the name of the header to add.
value is the CEL expression to apply to generate the output value for the header.
body controls manipulation of the HTTP body.
metadata stores CEL-evaluated values under the `metadata` CEL variable for subsequent policy evaluations. metadata is evaluated before header or body transformations.
Remove is a list of header names to remove from the request/response.
set is a list of headers and the value they should be set to.
the name of the header to add.
value is the CEL expression to apply to generate the output value for the header.
tunnel defines settings for managing tunnel connections (like HTTPS_PROXY) to the backend.
backendRef references the proxy server to reach. Supported types: Service and Backend.
Group is the group of the referent. For example, "gateway.networking.k8s.io". When unspecified or empty string, core API group is inferred.
Kind is the Kubernetes resource kind of the referent. For example "Service". Defaults to "Service" when not specified. ExternalName services can refer to CNAME DNS records that may live outside of the cluster and as such are difficult to reason about in terms of conformance. They also may not be safe to forward to (see CVE-2021-25740 for more information). Implementations SHOULD NOT support ExternalName Services. Support: Core (Services with a type other than ExternalName) Support: Implementation-specific (Services with type ExternalName)
Name is the name of the referent.
Namespace is the namespace of the backend. When unspecified, the local namespace is inferred. Note that when a namespace different than the local namespace is specified, a ReferenceGrant object is required in the referent namespace to allow that namespace's owner to accept the reference. See the ReferenceGrant documentation for details. Support: Core
Port specifies the destination port number to use for this resource. Port is required when the referent is a Kubernetes Service. In this case, the port number is the service port number, not the target port. For other resources, destination port might be derived from the referent resource or this field.
Regular expression (regex) matching for prompt guards and data masking.
The action to take if a regex pattern is matched in a request or response. This setting applies only to request matches. PromptguardResponse matches are always masked by default. Defaults to `Mask`.
A list of built-in regex patterns to match against the request or response. Matches and built-ins are additive.
A list of regex patterns to match against the request or response. Matches and built-ins are additive.
A custom response message to return to the client. If not specified, defaults to "The request was rejected due to inappropriate content".
A custom response message to return to the client. If not specified, defaults to "The request was rejected due to inappropriate content".
The status code to return to the client. Defaults to 403.
Configure a webhook to forward requests to for prompt guarding.
backendRef references the webhook server to reach. Supported types: Service and Backend.
Group is the group of the referent. For example, "gateway.networking.k8s.io". When unspecified or empty string, core API group is inferred.
Kind is the Kubernetes resource kind of the referent. For example "Service". Defaults to "Service" when not specified. ExternalName services can refer to CNAME DNS records that may live outside of the cluster and as such are difficult to reason about in terms of conformance. They also may not be safe to forward to (see CVE-2021-25740 for more information). Implementations SHOULD NOT support ExternalName Services. Support: Core (Services with a type other than ExternalName) Support: Implementation-specific (Services with type ExternalName)
Name is the name of the referent.
Namespace is the namespace of the backend. When unspecified, the local namespace is inferred. Note that when a namespace different than the local namespace is specified, a ReferenceGrant object is required in the referent namespace to allow that namespace's owner to accept the reference. See the ReferenceGrant documentation for details. Support: Core
Port specifies the destination port number to use for this resource. Port is required when the referent is a Kubernetes Service. In this case, the port number is the service port number, not the target port. For other resources, destination port might be derived from the referent resource or this field.
ForwardHeaderMatches defines a list of HTTP header matches that will be used to select the headers to forward to the webhook. Request headers are used when forwarding requests and response headers are used when forwarding responses. By default, no headers are forwarded.
Name is the name of the HTTP Header to be matched. Name matching MUST be case-insensitive. (See https://tools.ietf.org/html/rfc7230#section-3.2). If multiple entries specify equivalent header names, only the first entry with an equivalent name MUST be considered for a match. Subsequent entries with an equivalent header name MUST be ignored. Due to the case-insensitivity of header names, "foo" and "Foo" are considered equivalent. When a header is repeated in an HTTP request, it is implementation-specific behavior as to how this is represented. Generally, proxies should follow the guidance from the RFC: https://www.rfc-editor.org/rfc/rfc7230.html#section-3.2.2 regarding processing a repeated header, with special handling for "Set-Cookie".
Type specifies how to match against the value of the header. Support: Core (Exact) Support: Implementation-specific (RegularExpression) Since RegularExpression HeaderMatchType has implementation-specific conformance, implementations can support POSIX, PCRE or any other dialects of regular expressions. Please read the implementation's documentation to determine the supported dialect.
Value is the value of HTTP Header to be matched. <gateway:experimental:description> Must consist of printable US-ASCII characters, optionally separated by single tabs or spaces. See: https://tools.ietf.org/html/rfc7230#section-3.2 </gateway:experimental:description> <gateway:experimental:validation:Pattern=`^[!-~]+([\t ]?[!-~]+)*$`>
Prompt guards to apply to responses returned by the LLM provider.
bedrockGuardrails configures AWS Bedrock Guardrails for prompt guarding.
GuardrailIdentifier is the identifier of the Guardrail policy to use for the backend.
policies controls policies for communicating with AWS Bedrock Guardrails.
auth defines settings for managing authentication to the backend
Auth specifies an explicit AWS authentication method for the backend. When omitted, we will try to use the default AWS SDK authentication methods.
SecretRef references a Kubernetes Secret containing the AWS credentials. The Secret must have keys "accessKey", "secretKey", and optionally "sessionToken".
Name of the referent. This field is effectively required, but due to backwards compatibility is allowed to be empty. Instances of this type with an empty value here are almost certainly wrong. More info: https://kubernetes.io/docs/concepts/overview/working-with-objects/names/#names
Azure specifies an Azure authentication method for the backend.
Details for managed identity authentication
SecretRef references a Kubernetes Secret containing the Azure credentials. The Secret must have keys "clientId", "tenantId", and "clientSecret".
Name of the referent. This field is effectively required, but due to backwards compatibility is allowed to be empty. Instances of this type with an empty value here are almost certainly wrong. More info: https://kubernetes.io/docs/concepts/overview/working-with-objects/names/#names
Auth specifies to use a Google authentication method for the backend. When omitted, we will try to use the default AWS SDK authentication methods.
audience allows explicitly configuring the `aud` of the ID Token. Ony valid with `IdToken` type. If not set, the aud is automatically derived from the backend hostname.
The type of token to generate. To authenticate to GCP services, generally an AccessToken is used. To authenticate to CloudRun, an IdToken is used.
key provides an inline key to use as the value of the Authorization header. This option is the least secure; usage of a Secret is preferred.
passthrough passes through an existing token that has been sent by the client and validated. Other policies, like JWT and API Key authentication, will strip the original client credentials. Passthrough backend authentication causes the original token to be added back into the request. If there are no client authentication policies on the request, the original token would be unchanged, so this would have no effect.
secretRef references a Kubernetes secret storing the key to use the authorization value. This must be stored in the 'Authorization' key.
Name of the referent. This field is effectively required, but due to backwards compatibility is allowed to be empty. Instances of this type with an empty value here are almost certainly wrong. More info: https://kubernetes.io/docs/concepts/overview/working-with-objects/names/#names
health defines settings for passive and active health checking.
Eviction defines settings for evicting unhealthy backends.
ConsecutiveFailures is the number of consecutive unhealthy responses required before the backend is evicted. For example, a value of 5 means the backend must receive 5 unhealthy responses in a row before being evicted. When both consecutiveFailures and healthThreshold are set, the backend is evicted when either condition is met. When neither is set, a single unhealthy response can trigger eviction.
Duration specifies the base time a backend should be evicted after being marked unhealthy. Subsequent evictions use multiplicative backoff (duration * times_evicted). If all endpoints are evicted, the load balancer falls back to returning evicted endpoints rather than failing entirely. If unset, defaults to 3s.
HealthThreshold is the EWMA (exponentially-weighted moving average) health score threshold, expressed as 0–100. When set, a backend is only evicted if its computed health drops below this value after an unhealthy response. For example, 50 means the backend is evicted when its EWMA health falls below 50% following failures. Unlike consecutiveFailures (which counts consecutive failures), this uses a sliding-window average so a single success in a stream of failures can delay eviction. When both consecutiveFailures and healthThreshold are set, the backend is evicted when either condition is met. When neither is set, a single unhealthy response triggers eviction.
RestoreHealth is the health score (0–100) assigned to a backend when it returns from eviction. For gradual recovery, set below 100; for full recovery immediately, set 100. If unset, the backend resumes with the health it had when evicted.
UnhealthyCondition is a CEL expression that determines whether a response indicates an unhealthy backend. When the expression evaluates to true, the backend is considered unhealthy and may be evicted. For example, to evict on 5xx responses: `response.code >= 500`. When unset, any 5xx response, or a connection failure, is treated as unhealthy. This default lowers the backend's health score but does not trigger eviction on its own.
http defines settings for managing HTTP requests to the backend.
requestTimeout specifies the deadline for receiving a response from the backend.
version specifies the HTTP protocol version to use when connecting to the backend. If not specified, the version is automatically determined: * Service types can specify it with 'appProtocol' on the Service port. * If traffic is identified as gRPC, HTTP2 is used. * If the incoming traffic was plaintext HTTP, the original protocol will be used. * If the incoming traffic was HTTPS, HTTP1 will be used. This is because most clients will transparently upgrade HTTPS traffic to HTTP2, even if the backend doesn't support it
tcp defines settings for managing TCP connections to the backend.
connectTimeout defines the deadline for establishing a connection to the destination.
keepAlive defines settings for enabling TCP keepalives on the connection.
interval specifies the number of seconds between keep-alive probes. If unset, this defaults to 180s.
retries specifies the maximum number of keep-alive probes to send before dropping the connection. If unset, this defaults to 9.
time specifies the number of seconds a connection needs to be idle before keep-alive probes start being sent. If unset, this defaults to 180s.
tls defines settings for managing TLS connections to the backend. If this field is set, TLS will be initiated to the backend; the system trusted CA certificates will be used to validate the server, and the SNI will automatically be set based on the destination.
alpnProtocols sets the Application Level Protocol Negotiation (ALPN) value to use in the TLS handshake. If not present, defaults to ["h2", "http/1.1"].
caCertificateRefs defines the CA certificate ConfigMap to use to verify the server certificate. If unset, the system's trusted certificates are used.
Name of the referent. This field is effectively required, but due to backwards compatibility is allowed to be empty. Instances of this type with an empty value here are almost certainly wrong. More info: https://kubernetes.io/docs/concepts/overview/working-with-objects/names/#names
insecureSkipVerify originates TLS but skips verification of the backend's certificate. WARNING: This is an insecure option that should only be used if the risks are understood. There are two modes: * All disables all TLS verification * Hostname verifies the CA certificate is trusted, but ignores any mismatch of hostname/SANs. Note that this method is still insecure; prefer setting verifySubjectAltNames to customize the valid hostnames if possible.
mtlsCertificateRef enables mutual TLS to the backend, using the specified key (tls.key) and cert (tls.crt) from the refenced Secret. An optional 'ca.cert' field, if present, will be used to verify the server certificate if present. If caCertificateRefs is also specified, the caCertificateRefs field takes priority. If unspecified, no client certificate will be used.
Name of the referent. This field is effectively required, but due to backwards compatibility is allowed to be empty. Instances of this type with an empty value here are almost certainly wrong. More info: https://kubernetes.io/docs/concepts/overview/working-with-objects/names/#names
sni specifies the Server Name Indicator (SNI) to be used in the TLS handshake. If unset, the SNI is automatically set based on the destination hostname.
verifySubjectAltNames specifies the Subject Alternative Names (SAN) to verify in the server certificate. If not present, the destination hostname is automatically used.
transformation is used to mutate and transform requests and responses sent to and from the backend.
request is used to modify the request path.
add is a list of headers to add to the request and what that value should be set to. If there is already a header with these values then append the value as an extra entry.
the name of the header to add.
value is the CEL expression to apply to generate the output value for the header.
body controls manipulation of the HTTP body.
metadata stores CEL-evaluated values under the `metadata` CEL variable for subsequent policy evaluations. metadata is evaluated before header or body transformations.
Remove is a list of header names to remove from the request/response.
set is a list of headers and the value they should be set to.
the name of the header to add.
value is the CEL expression to apply to generate the output value for the header.
response is used to modify the response path.
add is a list of headers to add to the request and what that value should be set to. If there is already a header with these values then append the value as an extra entry.
the name of the header to add.
value is the CEL expression to apply to generate the output value for the header.
body controls manipulation of the HTTP body.
metadata stores CEL-evaluated values under the `metadata` CEL variable for subsequent policy evaluations. metadata is evaluated before header or body transformations.
Remove is a list of header names to remove from the request/response.
set is a list of headers and the value they should be set to.
the name of the header to add.
value is the CEL expression to apply to generate the output value for the header.
tunnel defines settings for managing tunnel connections (like HTTPS_PROXY) to the backend.
backendRef references the proxy server to reach. Supported types: Service and Backend.
Group is the group of the referent. For example, "gateway.networking.k8s.io". When unspecified or empty string, core API group is inferred.
Kind is the Kubernetes resource kind of the referent. For example "Service". Defaults to "Service" when not specified. ExternalName services can refer to CNAME DNS records that may live outside of the cluster and as such are difficult to reason about in terms of conformance. They also may not be safe to forward to (see CVE-2021-25740 for more information). Implementations SHOULD NOT support ExternalName Services. Support: Core (Services with a type other than ExternalName) Support: Implementation-specific (Services with type ExternalName)
Name is the name of the referent.
Namespace is the namespace of the backend. When unspecified, the local namespace is inferred. Note that when a namespace different than the local namespace is specified, a ReferenceGrant object is required in the referent namespace to allow that namespace's owner to accept the reference. See the ReferenceGrant documentation for details. Support: Core
Port specifies the destination port number to use for this resource. Port is required when the referent is a Kubernetes Service. In this case, the port number is the service port number, not the target port. For other resources, destination port might be derived from the referent resource or this field.
Region is the AWS region where the guardrail is deployed (e.g., "us-west-2").
GuardrailVersion is the version of the Guardrail policy to use for the backend.
googleModelArmor configures Google Model Armor for prompt guarding.
Location is the Google Cloud location (e.g., "us-central1"). Defaults to "us-central1" if not specified.
policies controls policies for communicating with Google Model Armor.
auth defines settings for managing authentication to the backend
Auth specifies an explicit AWS authentication method for the backend. When omitted, we will try to use the default AWS SDK authentication methods.
SecretRef references a Kubernetes Secret containing the AWS credentials. The Secret must have keys "accessKey", "secretKey", and optionally "sessionToken".
Name of the referent. This field is effectively required, but due to backwards compatibility is allowed to be empty. Instances of this type with an empty value here are almost certainly wrong. More info: https://kubernetes.io/docs/concepts/overview/working-with-objects/names/#names
Azure specifies an Azure authentication method for the backend.
Details for managed identity authentication
SecretRef references a Kubernetes Secret containing the Azure credentials. The Secret must have keys "clientId", "tenantId", and "clientSecret".
Name of the referent. This field is effectively required, but due to backwards compatibility is allowed to be empty. Instances of this type with an empty value here are almost certainly wrong. More info: https://kubernetes.io/docs/concepts/overview/working-with-objects/names/#names
Auth specifies to use a Google authentication method for the backend. When omitted, we will try to use the default AWS SDK authentication methods.
audience allows explicitly configuring the `aud` of the ID Token. Ony valid with `IdToken` type. If not set, the aud is automatically derived from the backend hostname.
The type of token to generate. To authenticate to GCP services, generally an AccessToken is used. To authenticate to CloudRun, an IdToken is used.
key provides an inline key to use as the value of the Authorization header. This option is the least secure; usage of a Secret is preferred.
passthrough passes through an existing token that has been sent by the client and validated. Other policies, like JWT and API Key authentication, will strip the original client credentials. Passthrough backend authentication causes the original token to be added back into the request. If there are no client authentication policies on the request, the original token would be unchanged, so this would have no effect.
secretRef references a Kubernetes secret storing the key to use the authorization value. This must be stored in the 'Authorization' key.
Name of the referent. This field is effectively required, but due to backwards compatibility is allowed to be empty. Instances of this type with an empty value here are almost certainly wrong. More info: https://kubernetes.io/docs/concepts/overview/working-with-objects/names/#names
health defines settings for passive and active health checking.
Eviction defines settings for evicting unhealthy backends.
ConsecutiveFailures is the number of consecutive unhealthy responses required before the backend is evicted. For example, a value of 5 means the backend must receive 5 unhealthy responses in a row before being evicted. When both consecutiveFailures and healthThreshold are set, the backend is evicted when either condition is met. When neither is set, a single unhealthy response can trigger eviction.
Duration specifies the base time a backend should be evicted after being marked unhealthy. Subsequent evictions use multiplicative backoff (duration * times_evicted). If all endpoints are evicted, the load balancer falls back to returning evicted endpoints rather than failing entirely. If unset, defaults to 3s.
HealthThreshold is the EWMA (exponentially-weighted moving average) health score threshold, expressed as 0–100. When set, a backend is only evicted if its computed health drops below this value after an unhealthy response. For example, 50 means the backend is evicted when its EWMA health falls below 50% following failures. Unlike consecutiveFailures (which counts consecutive failures), this uses a sliding-window average so a single success in a stream of failures can delay eviction. When both consecutiveFailures and healthThreshold are set, the backend is evicted when either condition is met. When neither is set, a single unhealthy response triggers eviction.
RestoreHealth is the health score (0–100) assigned to a backend when it returns from eviction. For gradual recovery, set below 100; for full recovery immediately, set 100. If unset, the backend resumes with the health it had when evicted.
UnhealthyCondition is a CEL expression that determines whether a response indicates an unhealthy backend. When the expression evaluates to true, the backend is considered unhealthy and may be evicted. For example, to evict on 5xx responses: `response.code >= 500`. When unset, any 5xx response, or a connection failure, is treated as unhealthy. This default lowers the backend's health score but does not trigger eviction on its own.
http defines settings for managing HTTP requests to the backend.
requestTimeout specifies the deadline for receiving a response from the backend.
version specifies the HTTP protocol version to use when connecting to the backend. If not specified, the version is automatically determined: * Service types can specify it with 'appProtocol' on the Service port. * If traffic is identified as gRPC, HTTP2 is used. * If the incoming traffic was plaintext HTTP, the original protocol will be used. * If the incoming traffic was HTTPS, HTTP1 will be used. This is because most clients will transparently upgrade HTTPS traffic to HTTP2, even if the backend doesn't support it
tcp defines settings for managing TCP connections to the backend.
connectTimeout defines the deadline for establishing a connection to the destination.
keepAlive defines settings for enabling TCP keepalives on the connection.
interval specifies the number of seconds between keep-alive probes. If unset, this defaults to 180s.
retries specifies the maximum number of keep-alive probes to send before dropping the connection. If unset, this defaults to 9.
time specifies the number of seconds a connection needs to be idle before keep-alive probes start being sent. If unset, this defaults to 180s.
tls defines settings for managing TLS connections to the backend. If this field is set, TLS will be initiated to the backend; the system trusted CA certificates will be used to validate the server, and the SNI will automatically be set based on the destination.
alpnProtocols sets the Application Level Protocol Negotiation (ALPN) value to use in the TLS handshake. If not present, defaults to ["h2", "http/1.1"].
caCertificateRefs defines the CA certificate ConfigMap to use to verify the server certificate. If unset, the system's trusted certificates are used.
Name of the referent. This field is effectively required, but due to backwards compatibility is allowed to be empty. Instances of this type with an empty value here are almost certainly wrong. More info: https://kubernetes.io/docs/concepts/overview/working-with-objects/names/#names
insecureSkipVerify originates TLS but skips verification of the backend's certificate. WARNING: This is an insecure option that should only be used if the risks are understood. There are two modes: * All disables all TLS verification * Hostname verifies the CA certificate is trusted, but ignores any mismatch of hostname/SANs. Note that this method is still insecure; prefer setting verifySubjectAltNames to customize the valid hostnames if possible.
mtlsCertificateRef enables mutual TLS to the backend, using the specified key (tls.key) and cert (tls.crt) from the refenced Secret. An optional 'ca.cert' field, if present, will be used to verify the server certificate if present. If caCertificateRefs is also specified, the caCertificateRefs field takes priority. If unspecified, no client certificate will be used.
Name of the referent. This field is effectively required, but due to backwards compatibility is allowed to be empty. Instances of this type with an empty value here are almost certainly wrong. More info: https://kubernetes.io/docs/concepts/overview/working-with-objects/names/#names
sni specifies the Server Name Indicator (SNI) to be used in the TLS handshake. If unset, the SNI is automatically set based on the destination hostname.
verifySubjectAltNames specifies the Subject Alternative Names (SAN) to verify in the server certificate. If not present, the destination hostname is automatically used.
transformation is used to mutate and transform requests and responses sent to and from the backend.
request is used to modify the request path.
add is a list of headers to add to the request and what that value should be set to. If there is already a header with these values then append the value as an extra entry.
the name of the header to add.
value is the CEL expression to apply to generate the output value for the header.
body controls manipulation of the HTTP body.
metadata stores CEL-evaluated values under the `metadata` CEL variable for subsequent policy evaluations. metadata is evaluated before header or body transformations.
Remove is a list of header names to remove from the request/response.
set is a list of headers and the value they should be set to.
the name of the header to add.
value is the CEL expression to apply to generate the output value for the header.
response is used to modify the response path.
add is a list of headers to add to the request and what that value should be set to. If there is already a header with these values then append the value as an extra entry.
the name of the header to add.
value is the CEL expression to apply to generate the output value for the header.
body controls manipulation of the HTTP body.
metadata stores CEL-evaluated values under the `metadata` CEL variable for subsequent policy evaluations. metadata is evaluated before header or body transformations.
Remove is a list of header names to remove from the request/response.
set is a list of headers and the value they should be set to.
the name of the header to add.
value is the CEL expression to apply to generate the output value for the header.
tunnel defines settings for managing tunnel connections (like HTTPS_PROXY) to the backend.
backendRef references the proxy server to reach. Supported types: Service and Backend.
Group is the group of the referent. For example, "gateway.networking.k8s.io". When unspecified or empty string, core API group is inferred.
Kind is the Kubernetes resource kind of the referent. For example "Service". Defaults to "Service" when not specified. ExternalName services can refer to CNAME DNS records that may live outside of the cluster and as such are difficult to reason about in terms of conformance. They also may not be safe to forward to (see CVE-2021-25740 for more information). Implementations SHOULD NOT support ExternalName Services. Support: Core (Services with a type other than ExternalName) Support: Implementation-specific (Services with type ExternalName)
Name is the name of the referent.
Namespace is the namespace of the backend. When unspecified, the local namespace is inferred. Note that when a namespace different than the local namespace is specified, a ReferenceGrant object is required in the referent namespace to allow that namespace's owner to accept the reference. See the ReferenceGrant documentation for details. Support: Core
Port specifies the destination port number to use for this resource. Port is required when the referent is a Kubernetes Service. In this case, the port number is the service port number, not the target port. For other resources, destination port might be derived from the referent resource or this field.
ProjectID is the Google Cloud project ID.
TemplateID is the template ID for Google Model Armor.
Regular expression (regex) matching for prompt guards and data masking.
The action to take if a regex pattern is matched in a request or response. This setting applies only to request matches. PromptguardResponse matches are always masked by default. Defaults to `Mask`.
A list of built-in regex patterns to match against the request or response. Matches and built-ins are additive.
A list of regex patterns to match against the request or response. Matches and built-ins are additive.
A custom response message to return to the client. If not specified, defaults to "The response was rejected due to inappropriate content".
A custom response message to return to the client. If not specified, defaults to "The request was rejected due to inappropriate content".
The status code to return to the client. Defaults to 403.
Configure a webhook to forward responses to for prompt guarding.
backendRef references the webhook server to reach. Supported types: Service and Backend.
Group is the group of the referent. For example, "gateway.networking.k8s.io". When unspecified or empty string, core API group is inferred.
Kind is the Kubernetes resource kind of the referent. For example "Service". Defaults to "Service" when not specified. ExternalName services can refer to CNAME DNS records that may live outside of the cluster and as such are difficult to reason about in terms of conformance. They also may not be safe to forward to (see CVE-2021-25740 for more information). Implementations SHOULD NOT support ExternalName Services. Support: Core (Services with a type other than ExternalName) Support: Implementation-specific (Services with type ExternalName)
Name is the name of the referent.
Namespace is the namespace of the backend. When unspecified, the local namespace is inferred. Note that when a namespace different than the local namespace is specified, a ReferenceGrant object is required in the referent namespace to allow that namespace's owner to accept the reference. See the ReferenceGrant documentation for details. Support: Core
Port specifies the destination port number to use for this resource. Port is required when the referent is a Kubernetes Service. In this case, the port number is the service port number, not the target port. For other resources, destination port might be derived from the referent resource or this field.
ForwardHeaderMatches defines a list of HTTP header matches that will be used to select the headers to forward to the webhook. Request headers are used when forwarding requests and response headers are used when forwarding responses. By default, no headers are forwarded.
Name is the name of the HTTP Header to be matched. Name matching MUST be case-insensitive. (See https://tools.ietf.org/html/rfc7230#section-3.2). If multiple entries specify equivalent header names, only the first entry with an equivalent name MUST be considered for a match. Subsequent entries with an equivalent header name MUST be ignored. Due to the case-insensitivity of header names, "foo" and "Foo" are considered equivalent. When a header is repeated in an HTTP request, it is implementation-specific behavior as to how this is represented. Generally, proxies should follow the guidance from the RFC: https://www.rfc-editor.org/rfc/rfc7230.html#section-3.2.2 regarding processing a repeated header, with special handling for "Set-Cookie".
Type specifies how to match against the value of the header. Support: Core (Exact) Support: Implementation-specific (RegularExpression) Since RegularExpression HeaderMatchType has implementation-specific conformance, implementations can support POSIX, PCRE or any other dialects of regular expressions. Please read the implementation's documentation to determine the supported dialect.
Value is the value of HTTP Header to be matched. <gateway:experimental:description> Must consist of printable US-ASCII characters, optionally separated by single tabs or spaces. See: https://tools.ietf.org/html/rfc7230#section-3.2 </gateway:experimental:description> <gateway:experimental:validation:Pattern=`^[!-~]+([\t ]?[!-~]+)*$`>
routes defines how to identify the type of traffic to handle. The keys are URL path suffixes matched using ends-with comparison (e.g., "/v1/chat/completions"). The special "*" wildcard matches any path. If not specified, all traffic defaults to "completions" type.
Provide CEL transformations to compute and set fields in the request body. The expression result overwrites any existing value for that field. This has a higher priority than 'overrides' if both are set for the same key.
CEL expression used to compute the field value.
The name of the field to set.
auth defines settings for managing authentication to the backend
Auth specifies an explicit AWS authentication method for the backend. When omitted, we will try to use the default AWS SDK authentication methods.
SecretRef references a Kubernetes Secret containing the AWS credentials. The Secret must have keys "accessKey", "secretKey", and optionally "sessionToken".
Name of the referent. This field is effectively required, but due to backwards compatibility is allowed to be empty. Instances of this type with an empty value here are almost certainly wrong. More info: https://kubernetes.io/docs/concepts/overview/working-with-objects/names/#names
Azure specifies an Azure authentication method for the backend.
Details for managed identity authentication
SecretRef references a Kubernetes Secret containing the Azure credentials. The Secret must have keys "clientId", "tenantId", and "clientSecret".
Name of the referent. This field is effectively required, but due to backwards compatibility is allowed to be empty. Instances of this type with an empty value here are almost certainly wrong. More info: https://kubernetes.io/docs/concepts/overview/working-with-objects/names/#names
Auth specifies to use a Google authentication method for the backend. When omitted, we will try to use the default AWS SDK authentication methods.
audience allows explicitly configuring the `aud` of the ID Token. Ony valid with `IdToken` type. If not set, the aud is automatically derived from the backend hostname.
The type of token to generate. To authenticate to GCP services, generally an AccessToken is used. To authenticate to CloudRun, an IdToken is used.
key provides an inline key to use as the value of the Authorization header. This option is the least secure; usage of a Secret is preferred.
passthrough passes through an existing token that has been sent by the client and validated. Other policies, like JWT and API Key authentication, will strip the original client credentials. Passthrough backend authentication causes the original token to be added back into the request. If there are no client authentication policies on the request, the original token would be unchanged, so this would have no effect.
secretRef references a Kubernetes secret storing the key to use the authorization value. This must be stored in the 'Authorization' key.
Name of the referent. This field is effectively required, but due to backwards compatibility is allowed to be empty. Instances of this type with an empty value here are almost certainly wrong. More info: https://kubernetes.io/docs/concepts/overview/working-with-objects/names/#names
health defines settings for passive and active health checking.
Eviction defines settings for evicting unhealthy backends.
ConsecutiveFailures is the number of consecutive unhealthy responses required before the backend is evicted. For example, a value of 5 means the backend must receive 5 unhealthy responses in a row before being evicted. When both consecutiveFailures and healthThreshold are set, the backend is evicted when either condition is met. When neither is set, a single unhealthy response can trigger eviction.
Duration specifies the base time a backend should be evicted after being marked unhealthy. Subsequent evictions use multiplicative backoff (duration * times_evicted). If all endpoints are evicted, the load balancer falls back to returning evicted endpoints rather than failing entirely. If unset, defaults to 3s.
HealthThreshold is the EWMA (exponentially-weighted moving average) health score threshold, expressed as 0–100. When set, a backend is only evicted if its computed health drops below this value after an unhealthy response. For example, 50 means the backend is evicted when its EWMA health falls below 50% following failures. Unlike consecutiveFailures (which counts consecutive failures), this uses a sliding-window average so a single success in a stream of failures can delay eviction. When both consecutiveFailures and healthThreshold are set, the backend is evicted when either condition is met. When neither is set, a single unhealthy response triggers eviction.
RestoreHealth is the health score (0–100) assigned to a backend when it returns from eviction. For gradual recovery, set below 100; for full recovery immediately, set 100. If unset, the backend resumes with the health it had when evicted.
UnhealthyCondition is a CEL expression that determines whether a response indicates an unhealthy backend. When the expression evaluates to true, the backend is considered unhealthy and may be evicted. For example, to evict on 5xx responses: `response.code >= 500`. When unset, any 5xx response, or a connection failure, is treated as unhealthy. This default lowers the backend's health score but does not trigger eviction on its own.
http defines settings for managing HTTP requests to the backend.
requestTimeout specifies the deadline for receiving a response from the backend.
version specifies the HTTP protocol version to use when connecting to the backend. If not specified, the version is automatically determined: * Service types can specify it with 'appProtocol' on the Service port. * If traffic is identified as gRPC, HTTP2 is used. * If the incoming traffic was plaintext HTTP, the original protocol will be used. * If the incoming traffic was HTTPS, HTTP1 will be used. This is because most clients will transparently upgrade HTTPS traffic to HTTP2, even if the backend doesn't support it
mcp specifies settings for MCP workloads. This is only applicable when connecting to a Backend of type 'mcp'.
authentication defines MCPBackend specific authentication rules.
audiences specify the list of allowed audiences that are allowed access. This corresponds to the 'aud' claim (https://datatracker.ietf.org/doc/html/rfc7519#section-4.1.3). If unset, any audience is allowed.
Issuer identifies the IdP that issued the JWT. This corresponds to the 'iss' claim (https://tools.ietf.org/html/rfc7519#section-4.1.1).
jwks defines the remote JSON Web Key used to validate the signature of the JWT.
backendRef references the remote JWKS server to reach. Supported types are Service and (static) Backend. An AgentgatewayPolicy containing backend tls config can then be attached to the service/backend in order to set tls options for a connection to the remote jwks source.
Group is the group of the referent. For example, "gateway.networking.k8s.io". When unspecified or empty string, core API group is inferred.
Kind is the Kubernetes resource kind of the referent. For example "Service". Defaults to "Service" when not specified. ExternalName services can refer to CNAME DNS records that may live outside of the cluster and as such are difficult to reason about in terms of conformance. They also may not be safe to forward to (see CVE-2021-25740 for more information). Implementations SHOULD NOT support ExternalName Services. Support: Core (Services with a type other than ExternalName) Support: Implementation-specific (Services with type ExternalName)
Name is the name of the referent.
Namespace is the namespace of the backend. When unspecified, the local namespace is inferred. Note that when a namespace different than the local namespace is specified, a ReferenceGrant object is required in the referent namespace to allow that namespace's owner to accept the reference. See the ReferenceGrant documentation for details. Support: Core
Port specifies the destination port number to use for this resource. Port is required when the referent is a Kubernetes Service. In this case, the port number is the service port number, not the target port. For other resources, destination port might be derived from the referent resource or this field.
Path to IdP jwks endpoint, relative to the root, commonly ".well-known/jwks.json".
validation mode for JWT authentication.
McpIDP specifies the identity provider to use for authentication
ResourceMetadata defines the metadata to use for MCP resources.
authorization defines MCPBackend level authorization. Unlike authorization at the HTTP level, which will reject unauthorized requests with a 403 error, this policy works at the MCPBackend level. List operations, such as list_tools, will have each item evaluated. Items that do not meet the rule will be filtered. Get or call operations, such as call_tool, will evaluate the specific item and reject requests that do not meet the rule.
Action defines whether the rule allows or denies the request if matched. If unspecified, the default is "Allow".
Policy specifies the Authorization rule to evaluate. A policy matches when **any** of the conditions evaluates to true.
MatchExpressions defines a set of conditions that must be satisfied for the rule to match. These expression should be in the form of a Common Expression Language (CEL) expression.
tcp defines settings for managing TCP connections to the backend.
connectTimeout defines the deadline for establishing a connection to the destination.
keepAlive defines settings for enabling TCP keepalives on the connection.
interval specifies the number of seconds between keep-alive probes. If unset, this defaults to 180s.
retries specifies the maximum number of keep-alive probes to send before dropping the connection. If unset, this defaults to 9.
time specifies the number of seconds a connection needs to be idle before keep-alive probes start being sent. If unset, this defaults to 180s.
tls defines settings for managing TLS connections to the backend. If this field is set, TLS will be initiated to the backend; the system trusted CA certificates will be used to validate the server, and the SNI will automatically be set based on the destination.
alpnProtocols sets the Application Level Protocol Negotiation (ALPN) value to use in the TLS handshake. If not present, defaults to ["h2", "http/1.1"].
caCertificateRefs defines the CA certificate ConfigMap to use to verify the server certificate. If unset, the system's trusted certificates are used.
Name of the referent. This field is effectively required, but due to backwards compatibility is allowed to be empty. Instances of this type with an empty value here are almost certainly wrong. More info: https://kubernetes.io/docs/concepts/overview/working-with-objects/names/#names
insecureSkipVerify originates TLS but skips verification of the backend's certificate. WARNING: This is an insecure option that should only be used if the risks are understood. There are two modes: * All disables all TLS verification * Hostname verifies the CA certificate is trusted, but ignores any mismatch of hostname/SANs. Note that this method is still insecure; prefer setting verifySubjectAltNames to customize the valid hostnames if possible.
mtlsCertificateRef enables mutual TLS to the backend, using the specified key (tls.key) and cert (tls.crt) from the refenced Secret. An optional 'ca.cert' field, if present, will be used to verify the server certificate if present. If caCertificateRefs is also specified, the caCertificateRefs field takes priority. If unspecified, no client certificate will be used.
Name of the referent. This field is effectively required, but due to backwards compatibility is allowed to be empty. Instances of this type with an empty value here are almost certainly wrong. More info: https://kubernetes.io/docs/concepts/overview/working-with-objects/names/#names
sni specifies the Server Name Indicator (SNI) to be used in the TLS handshake. If unset, the SNI is automatically set based on the destination hostname.
verifySubjectAltNames specifies the Subject Alternative Names (SAN) to verify in the server certificate. If not present, the destination hostname is automatically used.
transformation is used to mutate and transform requests and responses sent to and from the backend.
request is used to modify the request path.
add is a list of headers to add to the request and what that value should be set to. If there is already a header with these values then append the value as an extra entry.
the name of the header to add.
value is the CEL expression to apply to generate the output value for the header.
body controls manipulation of the HTTP body.
metadata stores CEL-evaluated values under the `metadata` CEL variable for subsequent policy evaluations. metadata is evaluated before header or body transformations.
Remove is a list of header names to remove from the request/response.
set is a list of headers and the value they should be set to.
the name of the header to add.
value is the CEL expression to apply to generate the output value for the header.
response is used to modify the response path.
add is a list of headers to add to the request and what that value should be set to. If there is already a header with these values then append the value as an extra entry.
the name of the header to add.
value is the CEL expression to apply to generate the output value for the header.
body controls manipulation of the HTTP body.
metadata stores CEL-evaluated values under the `metadata` CEL variable for subsequent policy evaluations. metadata is evaluated before header or body transformations.
Remove is a list of header names to remove from the request/response.
set is a list of headers and the value they should be set to.
the name of the header to add.
value is the CEL expression to apply to generate the output value for the header.
tunnel defines settings for managing tunnel connections (like HTTPS_PROXY) to the backend.
backendRef references the proxy server to reach. Supported types: Service and Backend.
Group is the group of the referent. For example, "gateway.networking.k8s.io". When unspecified or empty string, core API group is inferred.
Kind is the Kubernetes resource kind of the referent. For example "Service". Defaults to "Service" when not specified. ExternalName services can refer to CNAME DNS records that may live outside of the cluster and as such are difficult to reason about in terms of conformance. They also may not be safe to forward to (see CVE-2021-25740 for more information). Implementations SHOULD NOT support ExternalName Services. Support: Core (Services with a type other than ExternalName) Support: Implementation-specific (Services with type ExternalName)
Name is the name of the referent.
Namespace is the namespace of the backend. When unspecified, the local namespace is inferred. Note that when a namespace different than the local namespace is specified, a ReferenceGrant object is required in the referent namespace to allow that namespace's owner to accept the reference. See the ReferenceGrant documentation for details. Support: Core
Port specifies the destination port number to use for this resource. Port is required when the referent is a Kubernetes Service. In this case, the port number is the service port number, not the target port. For other resources, destination port might be derived from the referent resource or this field.
static represents a static hostname.
host to connect to.
port to connect to.
status defines the current state of AgentgatewayBackend.
Conditions is the list of conditions for the backend.
lastTransitionTime is the last time the condition transitioned from one status to another. This should be when the underlying condition changed. If that is not known, then using the time when the API field changed is acceptable.
message is a human readable message indicating details about the transition. This may be an empty string.
observedGeneration represents the .metadata.generation that the condition was set based upon. For instance, if .metadata.generation is currently 12, but the .status.conditions[x].observedGeneration is 9, the condition is out of date with respect to the current state of the instance.
reason contains a programmatic identifier indicating the reason for the condition's last transition. Producers of specific condition types may define expected values and meanings for this field, and whether the values are considered a guaranteed API. The value should be a CamelCase string. This field may not be empty.
status of the condition, one of True, False, Unknown.
type of condition in CamelCase or in foo.example.com/CamelCase.
Source: agentgateway.dev_agentgatewaybackends.yaml
Changes to AgentgatewayBackend (agentgateway.dev/v1alpha1) in each Agentgateway version, compared to the version before it.
1 property has been added
Released · View AgentgatewayBackend in v1.0.1
AgentgatewayBackend was already present in this version of Agentgateway, the oldest one on kubespec.dev
Released · View AgentgatewayBackend in v1.0.0
No examples of AgentgatewayBackend yet. Contribute one on GitHub.
kubespec.dev