# Tetrate Agent Router documentation

> Enterprise AI gateway for routing, governance, and operations across AI models, agents, and MCP servers.

This file contains links to documentation sections following the llmstxt.org standard.

## Table of Contents

- [AI gateway for agents and MCP](https://docs.tetrate.ai/welcome.md): Route, govern, and observe every AI request through one gateway. Evaluate Agent Router on the Fully Managed tier in an afternoon: routing, budgets,...
- [AI capabilities](https://docs.tetrate.ai/product-architecture/ai-capabilities.md): How Agent Router passes structured outputs, multimodal inputs, and function calling through to any provider, translating across providers where map...
- [Architecture overview](https://docs.tetrate.ai/product-architecture/architecture-overview.md): How Tetrate Agent Router is structured: a customer-managed data plane and a Tetrate-hosted management plane that keep prompt and response data on-p...
- [Data flows](https://docs.tetrate.ai/product-architecture/data-flows.md): The three data flows in Tetrate Agent Router, explained from first principles: outbound configuration polling, the prompt and response request path...
- [Deployment models](https://docs.tetrate.ai/product-architecture/deployment-models.md): Where Agent Router runs: fully hosted options (Service and Enterprise Fully Managed) and Enterprise Self-Hosted Data Plane, where you install only ...
- [Guardrails](https://docs.tetrate.ai/product-architecture/guardrails.md): How the Guardrails dynamic module inspects AI traffic inline at the Tetrate Agent Router gateway: interception, evaluation, enforcement, and egress...
- [Key concepts](https://docs.tetrate.ai/product-architecture/key-concepts.md): The core vocabulary of Tetrate Agent Router: data and management planes, projects, fallback policies, traffic splitting, guardrails, MCP profiles, ...
- [Network and Security](https://docs.tetrate.ai/product-architecture/network-and-security.md): Network topology and security controls in Tetrate Agent Router: intranet deployment, private provider connectivity, fallback-based network isolatio...
- [Operations and Reliability](https://docs.tetrate.ai/product-architecture/operations-and-reliability.md): How a Tetrate Agent Router data plane stays available: control-plane resilience, multi-region deployment, health checks, autoscaling, infrastructur...
- [Planes and core components](https://docs.tetrate.ai/product-architecture/planes-and-components.md): The split-plane model behind Tetrate Agent Router: a Tetrate-hosted management plane, a customer-managed data plane, the outbound-only link between...
- [Cost Control and Budgets](https://docs.tetrate.ai/product-architecture/product-evaluation/cost-control-and-budgets.md): Graded evaluation scenarios for cost visibility, budgets, chargeback, rate limits, and leaked-key containment.
- [Enterprise Operations and Compliance](https://docs.tetrate.ai/product-architecture/product-evaluation/enterprise-operations-and-compliance.md): Evaluation scenarios for data residency, log retention, user communication, HA and DR, regional load balancing, and configuration as code.
- [Evaluation Scenarios](https://docs.tetrate.ai/product-architecture/product-evaluation/evaluation-scenarios.md): Hands-on evaluation scenarios for Tetrate Agent Router: graded, verifiable exercises that follow the installation and guide documentation end to end.
- [First Requests and Developer Onboarding](https://docs.tetrate.ai/product-architecture/product-evaluation/first-requests-and-developer-onboarding.md): Evaluation scenarios for the first API call, the sample chat app, the Playground, SDK drop-in, and coding agents.
- [Identity and Developer Governance](https://docs.tetrate.ai/product-architecture/product-evaluation/identity-and-developer-governance.md): Evaluation scenarios for single sign-on, IdP group mapping, and developer onboarding with key issuance and revocation.
- [Onboarding](https://docs.tetrate.ai/product-architecture/product-evaluation/installation-and-onboarding.md): Evaluation scenarios for installing the self-hosted data plane and gateway, completing onboarding, and configuring SSO.
- [MCP Gateway](https://docs.tetrate.ai/product-architecture/product-evaluation/mcp-gateway.md): Evaluation scenarios for MCP profiles, server governance, secrets and identity, and project scoping.
- [Models, Providers, and Projects](https://docs.tetrate.ai/product-architecture/product-evaluation/models-providers-and-projects.md): Evaluation scenarios for the model catalog, provider provisioning, project isolation, and data plane and gateway lifecycle.
- [Observability and Audit](https://docs.tetrate.ai/product-architecture/product-evaluation/observability-and-audit.md): Evaluation scenarios for traffic monitoring, OpenTelemetry export, request logging modes, audit trails, and SIEM export.
- [Routing, Resilience, and Provider Abstraction](https://docs.tetrate.ai/product-architecture/product-evaluation/routing-resilience-and-provider-abstraction.md): Evaluation scenarios for multi-provider routing, fallbacks, traffic splitting, advanced rules, BYOK, embeddings, and batch jobs.
- [Safety and Data Protection](https://docs.tetrate.ai/product-architecture/product-evaluation/safety-and-data-protection.md): Evaluation scenarios for guardrails, PII redaction, prompt injection defense, and coding assistant protection.
- [Product overview](https://docs.tetrate.ai/product-architecture/product-overview.md): What Tetrate Agent Router Enterprise is: one governed path between the teams, agents, and applications that consume AI and the providers that suppl...
- [Agent Router Service](https://docs.tetrate.ai/get-started/agent-router-service/index.md): Quick start for individual developers on the Agent Router Service tier: sign in, create an API key, and send a first routed request.
- [Dev Console quickstart](https://docs.tetrate.ai/get-started/agent-router-service/quickstart-console.md): A 10-minute walkthrough from sign-in to a first routed AI request: issue an API key, get the proxy endpoint, call a model, and verify it in Request...
- [Agent Router Service quickstart](https://docs.tetrate.ai/get-started/agent-router-service/quickstart-explore.md): A 10-minute self-serve walkthrough from sign-up to a first routed AI request on the Agent Router Service: create an account, add an API key, point ...
- [Set up Auth0 for evaluation](https://docs.tetrate.ai/get-started/auth0-for-evaluation.md): Ground-up Auth0 path for Fully Managed evaluations: tenant, Redirect URI from the wizard, Auth0 app, paste SSO, finish onboarding, confirm sign-in.
- [Evaluate Agent Router](https://docs.tetrate.ai/get-started/evaluation.md): Two ways to evaluate Tetrate Agent Router, depending on how the decision is being made. Run the fast-track first: it stands up a working environmen...
- [Fast track evaluation](https://docs.tetrate.ai/get-started/fully-managed/fast-track.md): Evaluate Agent Router on the Fully Managed tier in an afternoon: sign in, route your own AI traffic through the gateway, and read back exactly what...
- [Agent Router Enterprise Fully Managed](https://docs.tetrate.ai/get-started/fully-managed/index.md): Set up SSO, model catalog, users, and monitor token consumption while Tetrate manages the management and data planes.
- [Fully Managed onboarding](https://docs.tetrate.ai/get-started/fully-managed/onboarding.md): Onboard a Fully Managed management plane: organization details, SSO, LLM providers, and models. Tetrate operates the data plane, so no data plane c...
- [Admin Dashboard quickstart](https://docs.tetrate.ai/get-started/fully-managed/quickstart-admin-dashboard.md): A 10-minute walkthrough of essential Admin Dashboard tasks: the model catalog, provider connections, user accounts, and organization-wide usage and...
- [Data plane installation](https://docs.tetrate.ai/get-started/self-hosted-data-plane/data-plane-installation/index.md): Install the Agent Router data plane, the Controller and gateway, into a Kubernetes cluster and connect it to the management plane.
- [Installation guide for AWS](https://docs.tetrate.ai/get-started/self-hosted-data-plane/data-plane-installation/installation-guide-aws.md): Install the Agent Router data plane on AWS Elastic Kubernetes Service (EKS), from cluster prerequisites through a connected, running gateway.
- [Installation Guide for Azure using Helm](https://docs.tetrate.ai/get-started/self-hosted-data-plane/data-plane-installation/installation-guide-az-helm.md): Install the Agent Router data plane on Azure Kubernetes Service (AKS) using Helm and a private container registry.
- [Installation guide for Azure](https://docs.tetrate.ai/get-started/self-hosted-data-plane/data-plane-installation/installation-guide-az.md): Install the Agent Router data plane on Azure Kubernetes Service (AKS), from prerequisites through a connected, running gateway.
- [Installation guide for GCP](https://docs.tetrate.ai/get-started/self-hosted-data-plane/data-plane-installation/installation-guide-gcp.md): Install the Agent Router data plane on Google Kubernetes Engine (GKE) on Google Cloud Platform, from prerequisites to a running gateway.
- [Installation guide for Helm](https://docs.tetrate.ai/get-started/self-hosted-data-plane/data-plane-installation/installation-guide-helm.md): Install the Agent Router data plane with Helm for teams that own the release lifecycle and run a private container registry.
- [Retrieve data plane resources](https://docs.tetrate.ai/get-started/self-hosted-data-plane/data-plane-installation/k8s-resources-for-gitops.md): Render the Kubernetes manifests that tare install would deploy, without applying them, for review, dry runs, or GitOps commits.
- [Gateway installation guide for AKS](https://docs.tetrate.ai/get-started/self-hosted-data-plane/gateway-installation/gateway-installation-guide-az.md): Install the Agent Router gateway on Azure Kubernetes Service (AKS) with the tare CLI, terminating TLS and routing traffic to the data plane.
- [Gateway installation guide for GCP](https://docs.tetrate.ai/get-started/self-hosted-data-plane/gateway-installation/gateway-installation-guide-gcp.md): Install the Agent Router gateway on Google Cloud with the tare CLI, terminating TLS and routing traffic to the data plane.
- [Gateway Installation](https://docs.tetrate.ai/get-started/self-hosted-data-plane/gateway-installation/gateway-installation-guide.md): How the tare install-gateway command exposes and manages AI traffic to the data plane, with Agent Router–specific runbooks that follow.
- [Gateway Installation](https://docs.tetrate.ai/get-started/self-hosted-data-plane/gateway-installation/index.md): Install the Agent Router gateway into a self-hosted data plane and wire DNS and TLS, with cloud-specific variants for GCP and Azure.
- [Agent Router Enterprise Self-Hosted Data Plane](https://docs.tetrate.ai/get-started/self-hosted-data-plane/index.md): Set up the data plane in your Kubernetes cluster while Tetrate manages the management plane.
- [Self-Hosted onboarding](https://docs.tetrate.ai/get-started/self-hosted-data-plane/onboarding.md): Onboard a Self-Hosted Data Plane management plane: organization details, SSO, providers, models, and the data plane credential used to install the ...
- [Prerequisites](https://docs.tetrate.ai/get-started/self-hosted-data-plane/prerequisites.md): The access and environment requirements to verify before using Tetrate Agent Router, covering what developers and administrators each need from day...
- [Configure SSO](https://docs.tetrate.ai/get-started/self-hosted-data-plane/sso-configuration-guide.md): Connect Developer Console access to an OpenID Connect (OIDC) identity provider; single sign-on is required, with no password login.
- [Configure SSO role mapping](https://docs.tetrate.ai/get-started/sso-role-mapping-guide.md): Map OIDC identity-provider claims to Agent Router roles so directory changes propagate automatically on each login, with no manual work.
- [Where to begin](https://docs.tetrate.ai/get-started/where-to-begin.md): Orientation for the Get started section: the readers the documentation serves, the three deployment paths, and the right starting point for each.
- [Aggregate MCP servers into a profile](https://docs.tetrate.ai/guides/dev/aggregate-mcp-servers-into-a-profile.md): Combine multiple Model Context Protocol (MCP) servers into one profile, a single endpoint that gives AI clients access to many tools at once.
- [Level 1: First insights in ten minutes](https://docs.tetrate.ai/guides/dev/analytics-onboarding/first-insights-in-ten-minutes.md): The first level of the usage analytics onboarding path: send a burst of traffic and explain its cost, tokens, and latency inside ten timed minutes.
- [Usage Analytics Onboarding](https://docs.tetrate.ai/guides/dev/analytics-onboarding/index.md): A three-level, timed onboarding path through usage analytics: first insights in ten minutes, the shadow bill, and an agent efficiency scorecard.
- [Level 3: Profile agent efficiency](https://docs.tetrate.ai/guides/dev/analytics-onboarding/profile-agent-efficiency.md): The final level of the usage analytics onboarding path: token splits, cached-request share, and cost per task, distilled into a baseline scorecard ...
- [Level 2: Read the shadow bill](https://docs.tetrate.ai/guides/dev/analytics-onboarding/read-the-shadow-bill.md): The second level of the usage analytics onboarding path: put a dollar figure on what subscription traffic would cost at API list prices, in fifteen...
- [Apply advanced routing rules](https://docs.tetrate.ai/guides/dev/apply-advanced-routing-rules.md): Combine fallback policies and traffic splitting to keep AI requests succeeding when a provider fails and distribute traffic across backends.
- [Claude Code](https://docs.tetrate.ai/guides/dev/coding-agents/claude-code.md): Configure Claude Code to route through Agent Router so it uses the right API patterns, fallback routing, and cost tracking out of the box.
- [Codex CLI](https://docs.tetrate.ai/guides/dev/coding-agents/codex.md): Configure OpenAI's Codex CLI to route through Agent Router, using the right API endpoint, auth patterns, and available models.
- [Cursor CLI](https://docs.tetrate.ai/guides/dev/coding-agents/cursor.md): Configure Cursor to route through Agent Router so it generates the right API patterns and uses Agent Router features and best practices.
- [Set up Coding Agents & Tools](https://docs.tetrate.ai/guides/dev/coding-agents/index.md): Connect coding agents to Agent Router with ready-to-paste config: API patterns, model routing, fallback, and cost tracking for each tool.
- [Lovable](https://docs.tetrate.ai/guides/dev/coding-agents/lovable.md): Give Lovable project context about Agent Router so apps it generates call the gateway with the right base URL, auth, and endpoints.
- [Export telemetry to observability](https://docs.tetrate.ai/guides/dev/export-telemetry-to-an-observability-stack.md): Export Agent Router telemetry over OpenTelemetry to Grafana, Datadog, or New Relic for the deeper questions Request Logs alone cannot answer.
- [Generate embeddings](https://docs.tetrate.ai/guides/dev/generate-embeddings.md): Generate embeddings through Agent Router for semantic search, RAG, clustering, and recommendation, routed and tracked like any other request.
- [Improve resilience with fallbacks](https://docs.tetrate.ai/guides/dev/improve-resilience-with-fallbacks.md): Configure fallback chains so the gateway reroutes to a backup provider on errors, keeping AI traffic available through outages and rate limits.
- [Guides for Developers](https://docs.tetrate.ai/guides/dev/index.md): Goal-oriented guides for developers consuming AI models through the gateway from the Developer Console: routing, fallbacks, guardrails, MCP, and te...
- [Integrate the gateway with an app](https://docs.tetrate.ai/guides/dev/integrate-the-gateway-with-an-app.md): Connect an application to the Agent Router gateway by swapping a provider base URL and key, so requests flow through routing, fallback, and guardra...
- [Monitor traffic and usage](https://docs.tetrate.ai/guides/dev/monitor-traffic-and-usage.md): Monitor AI traffic, cost, and latency in Agent Router with Request Logs and Usage Analytics to catch bill spikes and regressions early.
- [Protect requests with guardrails](https://docs.tetrate.ai/guides/dev/protect-requests-with-guardrails.md): Apply guardrails in the gateway to redact PII and block unsafe content inline, protecting every request that routes through a protected API key.
- [Build a chat app](https://docs.tetrate.ai/guides/dev/quickstart/build-a-chat-app.md): Build a streaming chat app on Agent Router with multi-turn conversations, streaming responses, and automatic failover across providers.
- [Quickstarts](https://docs.tetrate.ai/guides/dev/quickstart/index.md): The shortest path from an Agent Router API key to working code: make a first API call, then build a chat app on the gateway.
- [Make an AI API call](https://docs.tetrate.ai/guides/dev/quickstart/make-an-api-call.md): Make a first AI API call through Agent Router: one OpenAI-compatible endpoint for every model, with automatic failover and usage tracking.
- [Reduce cost with traffic splitting](https://docs.tetrate.ai/guides/dev/reduce-cost-with-traffic-splitting.md): Use traffic splitting to send a share of requests to cheaper models, cutting AI cost without dropping the capability that workloads need.
- [Route requests across providers](https://docs.tetrate.ai/guides/dev/route-requests-across-providers.md): Route AI requests across multiple providers through one gateway, with a primary model, reserves, and cheaper or self-hosted backends.
- [Run batch and long-running jobs](https://docs.tetrate.ai/guides/dev/run-batch-and-long-running-jobs.md): Run batch and long-running AI jobs through Agent Router, from overnight scoring queues to large embedding runs and model evaluation suites.
- [Test prompts in the playground](https://docs.tetrate.ai/guides/dev/test-prompts-in-the-playground.md): Iterate on prompts in the Agent Router Playground, an interactive console for testing model routing and outputs without redeploying an app.
- [Use your own provider credentials](https://docs.tetrate.ai/guides/dev/use-your-own-provider-credentials.md): Bring your own provider credentials (BYOK) to Agent Router so each team or subsidiary routes through its own OpenAI, Anthropic, or cloud account.
- [Audit Agent Router activity](https://docs.tetrate.ai/guides/ops/audit-platform-activity.md): Answer who changed what and when with the Agent Router audit log, an immutable record of administrative actions across Agent Router.
- [Communicate with Agent Router users](https://docs.tetrate.ai/guides/ops/communicate-with-platform-users.md): Notify Console users about new models, maintenance windows, and deprecations from Agent Router, so changes never land silently on teams.
- [Configure data residency](https://docs.tetrate.ai/guides/ops/compliance/configure-data-residency-and-no-retention.md): Enforce data residency and provider no-retention in Agent Router Enterprise so regulated traffic stays in-region and prompts are not retained upstr...
- [Export audit decisions to SIEM](https://docs.tetrate.ai/guides/ops/compliance/export-audit-and-policy-decisions-to-a-siem.md): Export Agent Router audit and guardrail policy decisions to a SIEM, recording what the gateway decided, for which request, and why.
- [Compliance and Governance](https://docs.tetrate.ai/guides/ops/compliance/index.md): Operator guides for compliance posture: data residency and no-retention, log retention and purge, and exporting audit and policy decisions to a SIEM.
- [Manage log retention and purge](https://docs.tetrate.ai/guides/ops/compliance/manage-log-retention-and-purge.md): Govern request-log retention in Agent Router: set how long prompt and response data is kept and purge records to meet compliance obligations.
- [Configure single sign-on](https://docs.tetrate.ai/guides/ops/configure-single-sign-on.md): Configure single sign-on for Agent Router against an OIDC identity provider, so Agent Router access follows central identity and group membership.
- [Bill AI spend back to the teams that incur it](https://docs.tetrate.ai/guides/ops/cost-and-quota/bill-ai-spend-back-to-teams.md): Produce per-user and per-team consumption statements in Agent Router and export them for chargeback and showback into a finance or billing system.
- [Catch spend drift before the invoice does](https://docs.tetrate.ai/guides/ops/cost-and-quota/catch-spend-drift-early.md): Run a regular usage review in Agent Router so slow cost creep, a chatty agent, or a drifting workload is caught in weeks, not on the monthly invoice.
- [Choose the right cost control for each workload](https://docs.tetrate.ai/guides/ops/cost-and-quota/choose-the-right-cost-control.md): Pick the right Agent Router cost control per workload: rate limits for flow control, budgets for spend caps, analytics for observation, and when ea...
- [Control request log data](https://docs.tetrate.ai/guides/ops/cost-and-quota/configuring-request-logs.md): Control what prompt and response data leaves the cluster in Agent Router request logs, keeping sensitive content inside your infrastructure.
- [Contain a leaked key before it drains the budget](https://docs.tetrate.ai/guides/ops/cost-and-quota/contain-a-leaked-key.md): The incident response for a leaked or runaway Agent Router credential: bound the damage with a rate limit, revoke the key, and pause the workload i...
- [Get alerted to cost spikes as they happen](https://docs.tetrate.ai/guides/ops/cost-and-quota/get-alerted-to-cost-spikes.md): Export Agent Router's per-request cost telemetry into the observability stack and wire alert rules, so anomalies page someone in minutes instead of...
- [Keep each team's spend inside an agreed ceiling](https://docs.tetrate.ai/guides/ops/cost-and-quota/keep-team-spend-inside-a-ceiling.md): Turn a spend ceiling agreed with finance into an enforced Agent Router budget: read the baseline, choose the scope, set the monthly limit, and deci...
- [Know what every app and project actually costs](https://docs.tetrate.ai/guides/ops/cost-and-quota/know-what-every-app-and-project-costs.md): Answer what each app and project costs in Agent Router with caller-set tags, then export grouped statements for finance and team leads.
- [Raise a spend ceiling without interrupting delivery](https://docs.tetrate.ai/guides/ops/cost-and-quota/raise-a-spend-ceiling.md): Adjust an Agent Router budget in place: raise the limit mid-period, pause enforcement during renegotiation, or retire an obsolete budget, with ever...
- [See budget exhaustion coming before it blocks a team](https://docs.tetrate.ai/guides/ops/cost-and-quota/see-budget-exhaustion-coming.md): Read budget burn in the Agent Router Admin Dashboard: the pooled meter, the per-member health view, and billed versus passthrough usage, in time to...
- [Stop runaway workloads before they burn the budget](https://docs.tetrate.ai/guides/ops/cost-and-quota/stop-runaway-workloads.md): Set token, request, and concurrency rate limits per model, key, user, and group in Agent Router, so a retry storm or fan-out burst hits a wall inst...
- [Govern MCP server access](https://docs.tetrate.ai/guides/ops/govern-mcp-server-access.md): Govern which Model Context Protocol (MCP) servers AI clients can reach through Agent Router, vetting and controlling tool access centrally.
- [Guides for Platform Operators](https://docs.tetrate.ai/guides/ops/index.md): Goal-oriented guides for operators managing models, providers, users, and Agent Router settings from the Admin Dashboard of Tetrate Agent Router.
- [Load-balance across regions](https://docs.tetrate.ai/guides/ops/load-balance-across-regional-deployments.md): Load-balance AI traffic across regional deployments of the same model in Agent Router for lower latency, higher capacity, and residency control.
- [Manage configuration as code](https://docs.tetrate.ai/guides/ops/manage-configuration-as-code.md): Manage Agent Router configuration as code, keeping policies, budgets, the model catalog, and keys in version control with reviewed changes.
- [Connect providers across clouds](https://docs.tetrate.ai/guides/ops/managing-models-and-providers/connect-provider-subscriptions-across-clouds.md): Connect multiple provider subscriptions across clouds and regions in Agent Router, consolidating existing vendor contracts behind one gateway.
- [Provision models and providers](https://docs.tetrate.ai/guides/ops/managing-models-and-providers/managing-models-and-providers.md): Provision models and providers in Agent Router: connect providers, enable models in the organisation catalog, and prepare them for project grants.
- [Custom and self-hosted models](https://docs.tetrate.ai/guides/ops/managing-models-and-providers/provision-custom-and-self-hosted-models.md): Provision custom, self-hosted, and embedding models in Agent Router alongside vendor models like GPT, Claude, and Gemini.
- [Provision models via API](https://docs.tetrate.ai/guides/ops/managing-models-and-providers/provision-models-via-api.md): Provision providers and models through the Catalog API: upsert provider, set credential, upsert model with pricing and capabilities, assign to a pr...
- [Provision AWS Bedrock models](https://docs.tetrate.ai/guides/ops/managing-models-and-providers/provisioning-aws-bedrock-models.md): Configure AWS Bedrock as an AI model provider in Tetrate Agent Router and make its models routable to developers.
- [Map Entra groups to functions](https://docs.tetrate.ai/guides/ops/map-entra-id-groups-to-business-functions.md): Map Microsoft Entra ID groups to Agent Router roles and cost centers, reusing existing directory structure to govern access and attribution.
- [Onboard developers and issue keys](https://docs.tetrate.ai/guides/ops/onboard-developers-and-issue-keys.md): Onboard developers to Agent Router and issue API keys at scale, granting access to the people who consume models through the gateway.
- [Plan HA and disaster recovery](https://docs.tetrate.ai/guides/ops/plan-high-availability-and-disaster-recovery.md): Plan high availability and disaster recovery for Agent Router, the shared gateway every AI request depends on, to limit downtime and data loss.
- [Create a project and grant models](https://docs.tetrate.ai/guides/ops/projects/create-a-project.md): Create an isolated project in the Tetrate Agent Router Admin Dashboard and grant it the providers and models its members are allowed to use.
- [Deploy and register a data plane](https://docs.tetrate.ai/guides/ops/projects/deploy-a-data-plane.md): Provision a data plane in Tetrate Agent Router, install its data plane credential, deploy it to a Kubernetes cluster, and connect it to the managem...
- [Project and data plane management](https://docs.tetrate.ai/guides/ops/projects/index.md): Operator guides for projects, gateways, and data planes in Tetrate Agent Router: create isolated projects, provision gateways, deploy data planes, ...
- [Issue a project-scoped API key](https://docs.tetrate.ai/guides/ops/projects/issue-a-project-api-key.md): Create an API key scoped to a single project in Tetrate Agent Router, so it works only against that project's gateway and cannot reach another proj...
- [Manage multiple gateways on a data plane](https://docs.tetrate.ai/guides/ops/projects/manage-gateways-on-a-data-plane.md): Review and operate the gateways a single Tetrate Agent Router data plane provisions and manages, including their URLs, owning projects, config vers...
- [Manage project members and access](https://docs.tetrate.ai/guides/ops/projects/manage-project-members.md): Add members to a project in Tetrate Agent Router, assign project roles, and keep project access separate from data-plane operations access.
- [Migrate a gateway or upgrade its runtime](https://docs.tetrate.ai/guides/ops/projects/migrate-a-gateway.md): Move a Tetrate Agent Router gateway to a different data plane, or upgrade its runtime in place, without changing the customer-facing URL or droppin...
- [Provision a gateway for a project](https://docs.tetrate.ai/guides/ops/projects/provision-a-gateway.md): Provision a gateway on a data plane and bind it to a project in Tetrate Agent Router, giving the project a stable customer-facing inference URL.
- [Grant MCP servers and profiles to a project](https://docs.tetrate.ai/guides/ops/projects/scope-mcp-to-a-project.md): Scope Model Context Protocol (MCP) servers and profiles to a project in Tetrate Agent Router, so they are reachable only through that project's gat...
- [Run multiple Agent Router instances](https://docs.tetrate.ai/guides/ops/run-multiple-platform-instances.md): Run multiple Agent Router instances for staging and production, or per region and business unit, with configuration kept consistent across them.
- [Configure custom guardrails](https://docs.tetrate.ai/guides/ops/safety/configure-custom-guardrails.md): Configure custom guardrails in Agent Router to catch the PII and content categories specific to your organization, beyond generic provider filters.
- [Configure vendor guardrails](https://docs.tetrate.ai/guides/ops/safety/configure-vendor-guardrails.md): Configure vendor guardrails in Agent Router to inspect, block, redact, or flag AI traffic inline using managed content-safety providers.
- [Detect and block prompt injection](https://docs.tetrate.ai/guides/ops/safety/detect-and-block-prompt-injection.md): Detect and block prompt injection in Agent Router, stopping smuggled instructions that make a model ignore its system prompt or leak data.
- [Detect and redact sensitive data](https://docs.tetrate.ai/guides/ops/safety/detect-and-redact-sensitive-data.md): Detect and redact sensitive data at the Agent Router gateway, catching PII, secrets, and card numbers in prompts before they leave.
- [Safety and Data Protection](https://docs.tetrate.ai/guides/ops/safety/index.md): Operator guides for guardrails and data protection: vendor and custom guardrails, PII redaction, prompt injection defense, and coding assistant pol...
- [Protect coding-assistant traffic](https://docs.tetrate.ai/guides/ops/safety/protect-coding-assistant-traffic.md): Protect coding-assistant traffic in Agent Router with guardrails on the OpenAI-compatible interface these tools share, guarding code and secrets.
- [Secure MCP secrets and identity](https://docs.tetrate.ai/guides/ops/secure-mcp-secrets-and-identity.md): Secure Model Context Protocol (MCP) access in Agent Router with secret stores and identity pass-through, proving who makes each tool call.
- [Guardrails best practices & FAQ](https://docs.tetrate.ai/best-practices/guardrails-best-practices-and-faq.md): Roll out, tune, and operate guardrails in Agent Router Enterprise: monitor mode, latency ordering, fail-open and fail-closed settings, investigatio...
- [Best Practices](https://docs.tetrate.ai/best-practices/index.md): Recommended patterns for running Agent Router Enterprise in production: resilience, cost control, safety, and configuration management.
- [Management API reference](https://docs.tetrate.ai/reference/api/index.md): <p className="tare-lead">
- [Audit log events](https://docs.tetrate.ai/reference/audit-log-events.md): Reference for Agent Router audit log events: the immutable, structured record of administrative actions that supports SOC 2 and similar frameworks.
- [Compliance](https://docs.tetrate.ai/reference/compliance.md): Reference for the compliance posture of Tetrate Agent Router: certifications and attestations, data-residency options, and upstream-provider data-r...
- [External links](https://docs.tetrate.ai/reference/external-links.md): A curated list of external references cited across the Tetrate Agent Router documentation, grouped by project and subject to upstream change.
- [Gateway behavior](https://docs.tetrate.ai/reference/gateway-behavior.md): Reference for Agent Router gateway behavior: normalized provider errors, correlation IDs, and consistent observability across every provider.
- [Glossary](https://docs.tetrate.ai/reference/glossary.md): One-line definitions of the Agent Router–specific vocabulary used across the Tetrate Agent Router documentation, with cross-references to marketing...
- [References](https://docs.tetrate.ai/reference/index.md): Reference material for Tetrate Agent Router: SDK and API reference, gateway behavior, telemetry and audit schemas, sizing, compliance, and the glos...
- [Integrations](https://docs.tetrate.ai/reference/integrations.md): A map of everything that connects to Agent Router: SDKs and frameworks, coding agents, guardrail vendors, observability backends, and MCP clients.
- [OpenTelemetry metrics](https://docs.tetrate.ai/reference/otel-metrics.md): Reference for Agent Router observability: OpenTelemetry traces over OTLP and Prometheus-compatible metrics emitted by each data plane component.
- [Release Notes](https://docs.tetrate.ai/reference/release-notes.md): Version history for Agent Router: data plane, the tare CLI, and gateway components, with the change categories tracked in each release.
- [SDK Reference](https://docs.tetrate.ai/reference/sdk/index.md): Use existing OpenAI and Anthropic SDKs against the gateway, or download a typed Agent Router client for Go, Python, or TypeScript.
- [SDK quickstart for curl](https://docs.tetrate.ai/reference/sdk/quickstart-curl.md): Make a first read-only identity (whoami) call against the gateway with nothing but curl and an Agent Router API key.
- [SDK quickstart for Go](https://docs.tetrate.ai/reference/sdk/quickstart-go.md): Install the typed Agent Router Go SDK from the local tarball and make a first read-only identity (whoami) call against the gateway.
- [SDK quickstart for Python](https://docs.tetrate.ai/reference/sdk/quickstart-python.md): Install the typed Agent Router Python SDK from the local sdist and make a first read-only identity (whoami) call against the gateway.
- [SDK quickstart for TypeScript](https://docs.tetrate.ai/reference/sdk/quickstart-typescript.md): Install the typed Agent Router TypeScript SDK from the local pack and make a first read-only identity (whoami) call against the gateway.
- [Sizing and scale](https://docs.tetrate.ai/reference/sizing-and-scale.md): Capacity reference for sizing a Tetrate Agent Router deployment against a scale target, covering workload dimensions and resource planning.
- [Gateway APIs](https://docs.tetrate.ai/reference/supported-apis.md): The OpenAI- and Anthropic-compatible gateway APIs Agent Router supports, including Chat Completions, Responses, Messages, and model discovery via /...
- [Supported Providers](https://docs.tetrate.ai/reference/supported-providers.md): The AI providers Agent Router connects to, how each provider is authenticated, and where to find the authoritative list of routable models.
- [`azure-gateway.json` schema (`tare gateway install --type azure`)](https://docs.tetrate.ai/reference/tare-cli/azure-gateway-config.md): Reference for the azure-gateway.json config file consumed by tare gateway install --type azure through the --config flag.
- [`tare gateway config init`](https://docs.tetrate.ai/reference/tare-cli/gateway-config-init.md): Reference for tare gateway config init, which generates a starter gateway config file from a serve domain and cloud project.
- [`tare gateway config lint`](https://docs.tetrate.ai/reference/tare-cli/gateway-config-lint.md): Reference for tare gateway config lint, which validates a gcp-gateway.json or azure-gateway.json file before installation.
- [`gcp-gateway.json` schema (`tare gateway install`)](https://docs.tetrate.ai/reference/tare-cli/gcp-gateway-config.md): Reference for the gcp-gateway.json config file consumed by tare gateway install --type gcp through the --config flag.
- [Agent Router CLI Reference](https://docs.tetrate.ai/reference/tare-cli/index.md): Reference for the tare CLI that deploys and manages the Agent Router data plane, covering its commands and gateway config file schemas.
- [`tare gateway install`](https://docs.tetrate.ai/reference/tare-cli/install-gateway.md): Reference for tare gateway install, which provisions the serve gateway resources that expose the Agent Router data plane.
- [`tare install`](https://docs.tetrate.ai/reference/tare-cli/install.md): Reference for tare install, which deploys the complete Agent Router data plane into a Kubernetes cluster in one command.
- [TARE CLI](https://docs.tetrate.ai/reference/tare-cli/tare-cli-reference.md): Command reference for the `tare` CLI, the single-command operator tool that installs, upgrades, and manages the Agent Router data plane.
- [`tare uninstall`](https://docs.tetrate.ai/reference/tare-cli/uninstall.md): Reference for tare uninstall, which tears down the Agent Router data plane from a Kubernetes cluster.
- [`tare upgrade`](https://docs.tetrate.ai/reference/tare-cli/upgrade.md): Reference for the`tare upgrade` command, which upgrades an installed Agent Router data plane to the current embedded version.
## Data plane (gateway)
- [Supported APIs](https://docs.tetrate.ai/reference/supported-apis.md): Gateway egress endpoints for inference traffic (`/v1/*`). Use for install smoke tests and app integration — not the management Catalog API at `/v1/catalog/*`.
## API reference (control plane)
- Per-method docs: append `.md` to any API reference page URL (e.g. `/reference/api/catalog/agentrouter-catalog-v-1-catalog-service-list-models.md`). Params and schemas included.
- Full REST surface: see OpenAPI spec below (management/control-plane APIs; not gateway `/v1/*` inference paths).
## Optional
- [llms-full.txt](https://docs.tetrate.ai/llms-full.txt): Full narrative docs concatenated (API endpoint pages excluded)
- [OpenAPI Specification](https://docs.tetrate.ai/openapi/agentrouter-openapi-latest.yaml): Combined OpenAPI 3.1 spec for Agent Router APIs