# ForceAI Gateway > Onboard LLM providers and create A2A agents in the ForceAI Gateway. - [ForceAI Gateway guides](https://gateway.forceaisecurity.com/index.md): Onboard LLM providers and create A2A-protocol agents in the ForceAI Gateway. - [Admin onboarding](https://gateway.forceaisecurity.com/admin-onboarding.md): A new administrator's start-to-finish path: add a model, add a router, build a governed agent, publish it, and scope it to a team. - [Onboard an existing provider](https://gateway.forceaisecurity.com/providers/onboard-existing-provider.md): Add a model for a provider ForceAI already supports, in the dashboard or one API call. No code. - [Add a new provider](https://gateway.forceaisecurity.com/providers/add-new-provider.md): Make an unsupported OpenAI-compatible provider a first-class option in ForceAI. A short, mechanical code change. - [A2A agents overview](https://gateway.forceaisecurity.com/agents/overview.md): Register an external agent as an A2A endpoint in the ForceAI Gateway, then govern and call it like any model. - [Build your own agent on ForceAI's smart-router (A2A)](https://gateway.forceaisecurity.com/agents/forceai-llm-a2a-agent.md): Wrap ForceAI's smart-router in a thin A2A server so you get a governed agent whose brain is ForceAI's own model routing. - [Run a single-shot ephemeral agent](https://gateway.forceaisecurity.com/agents/ephemeral-agent.md): Spawn a ForceAI-managed agent instance on demand, run one task, and let it hard-kill itself - all over the runtime API. - [Run an agent in an isolated sandbox](https://gateway.forceaisecurity.com/agents/isolated-sandbox.md): Execute an agent's code inside a locked-down OpenSandbox box while its model calls stay governed by the gateway, and pick that isolation per agent from the dashboard. - [Run a multi-agent orchestration](https://gateway.forceaisecurity.com/agents/agent-orchestrator.md): Submit one task; a planner splits it into a dependency graph, a router picks a handler per sub-task (an ephemeral runtime agent or one of your registered A2A agents), ForceAI runs them in order, and a synthesizer returns the answer. - [Snowflake Cortex Agent (A2A)](https://gateway.forceaisecurity.com/agents/snowflake-cortex.md): Register a Snowflake Cortex Agent in the ForceAI Gateway over A2A, with PAT or key-pair JWT auth. - [AWS Bedrock AgentCore (A2A)](https://gateway.forceaisecurity.com/agents/aws-agentcore.md): Register a Bedrock AgentCore runtime in the ForceAI Gateway over native A2A with AWS SigV4. - [Microsoft 365 Copilot (A2A)](https://gateway.forceaisecurity.com/agents/m365-copilot.md): Register a Microsoft 365 agent in the ForceAI Gateway over A2A with Entra OAuth. - [Microsoft 365 Tool Adapter](https://gateway.forceaisecurity.com/agents/m365-adapter.md): Let Copilot Studio and declarative agents call governed ForceAI capability as a tool, authenticated with a Microsoft Entra delegated token. - [Configure the skill store (S3 or minio)](https://gateway.forceaisecurity.com/skills/configure-s3-store.md): Connect an object store once so model-agnostic skills load their bodies on demand. Credentials are entered by an admin and encrypted at rest. - [Knowledge Catalog (OKF)](https://gateway.forceaisecurity.com/knowledge-catalog/okf-grounding.md): Publish domain knowledge as OKF bundles and have the gateway inject the concepts that match a turn into the model call, in real time, on an opted-in request. - [Set up SSO for the dashboard](https://gateway.forceaisecurity.com/sso/configure-sso.md): Wire an OIDC identity provider (Keycloak, Okta, Auth0, generic OIDC) into the ForceAI dashboard login, then map users to roles and access groups. - [UI Settings reference](https://gateway.forceaisecurity.com/admin/ui-settings.md): Every toggle on Admin Settings -> UI Settings: what it does, the underlying flag, and when to turn it on. - [Response caching](https://gateway.forceaisecurity.com/admin/response-caching.md): Serve a repeated request from Redis instead of calling the provider, cutting cost and latency. How it is wired in ForceAI, how to tune it, and how to see cache hits. - [Memory](https://gateway.forceaisecurity.com/admin/memory.md): A persistent user/team key-value store, plus an opt-in hook that injects the caller's memories into a request so any model recalls stored facts without resending them. - [Budgets](https://gateway.forceaisecurity.com/admin/budgets.md): Cap spend at the org, team, user, virtual key, or agent level. How budgets attach, how enforcement works, and how to set one up. - [Model Evaluations](https://gateway.forceaisecurity.com/admin/model-evaluations.md): Score a model against a suite of cases before anyone depends on it. Author a suite, run a bake-off, read the report, and turn on enforcement without causing an outage. - [Team member key permissions](https://gateway.forceaisecurity.com/admin/team-key-permissions.md): Why an invited team member cannot create a virtual key by default, how to grant it, and exactly what isolation holds between two members of the same team. - [Connector providers (MCP)](https://gateway.forceaisecurity.com/tools/mcp-connector.md): Connect third-party providers through the ForceAI Connector, restrict each one to access groups, and read the audit trail of which action every call reached. - [Search tools](https://gateway.forceaisecurity.com/tools/search-tools.md): Give agents and keys governed web search. Configure a provider once, call it at /v1/search/{name}, and grant access per key or team. - [Tool policies](https://gateway.forceaisecurity.com/tools/tool-policies.md): Govern the function and MCP tools that models and agents call: auto-discover them, then set an input/output trust policy that the gateway enforces as a prompt-injection defense. - [Chat completion](https://gateway.forceaisecurity.com/api-reference/chat-completion.md): OpenAI-compatible chat completion. Name a model, or a Smart Router for automatic selection. - [Add a model](https://gateway.forceaisecurity.com/api-reference/add-model.md): Add a model for a supported provider using forceai_params. - [Add a virtual key](https://gateway.forceaisecurity.com/api-reference/add-key.md): Mint a virtual key scoped to models, with a budget, rate limits, and expiry. - [Add an A2A agent](https://gateway.forceaisecurity.com/api-reference/add-agent.md): Register an external agent (Snowflake Cortex, Bedrock AgentCore, M365 Copilot) over A2A. ## OpenAPI Specs - [openapi](/openapi.yaml)