# AgentBouncer

> AgentBouncer verifies HTTP Message Signatures from AI agents and applies project-specific access policies.

## Documentation

- [Full documentation](https://agentbouncer.io/docs.md): Complete AgentBouncer documentation in Markdown.
- [Complete context document](https://agentbouncer.io/llms-full.txt): Full documentation intended for AI-agent context loading.
- [English documentation](https://agentbouncer.io/en/docs): Human-readable English documentation.
- [Russian documentation](https://agentbouncer.io/ru/docs): Human-readable Russian documentation.
- [Spanish documentation](https://agentbouncer.io/es/docs): Human-readable Spanish documentation.
- [English Markdown](https://agentbouncer.io/en/docs.md): English documentation in Markdown.
- [Russian Markdown](https://agentbouncer.io/ru/docs.md): Russian documentation in Markdown.
- [Spanish Markdown](https://agentbouncer.io/es/docs.md): Spanish documentation in Markdown.

## Integration guides

- [Verification API](https://agentbouncer.io/en/docs#rest-api): REST API integration.
- [JavaScript SDK](https://agentbouncer.io/en/docs#sdk): JavaScript and TypeScript SDK.
- [Next.js integration](https://agentbouncer.io/en/docs#nextjs): Protect Next.js App Router endpoints.
- [MCP integration](https://agentbouncer.io/en/docs#mcp): Protect MCP tools.
- [Project signing keys](https://agentbouncer.io/en/docs#project-signing-keys): Manage project-owned agent identities.
- [Providers](https://agentbouncer.io/en/docs#providers): Publish provider public-key directories.
- [Policies](https://agentbouncer.io/en/docs#policies): Configure authorization policies.
- [Verification response](https://agentbouncer.io/en/docs#response): Response object reference.
- [Errors](https://agentbouncer.io/en/docs#errors): Verification and policy reasons.
- [Troubleshooting](https://agentbouncer.io/en/docs#troubleshooting): Diagnose integration problems.

## API

- [POST /api/v1/verify](https://agentbouncer.io/api/v1/verify): Verify an incoming signed agent request.

## Important authorization rule

Use the `allowed` field as the final authorization decision.

The `verified` field reports cryptographic identity verification. The `allowed` field reports the final authorization result after applying the active project policy.