# usezn

> usezn is an AI security gateway for MCP, A2A, and HTTP agents. It checks untrusted tool-call input before execution and returns allow/block decisions with evidence.

## Canonical resources

- [Developer portal](https://usezn.com/developers/) - API keys, REST quickstart, MCP, CLI, and sandbox
- [Documentation](https://usezn.com/docs/) - full integration guide
- [OpenAPI](https://usezn.com/openapi.json) - machine-readable API schema
- [Agent guide](https://usezn.com/agents.md) - integration pattern and verdict handling
- [MCP manifest](https://usezn.com/.well-known/mcp.json) - Streamable HTTP endpoint and tools
- [Sitemap](https://usezn.com/sitemap.xml) - public URL inventory
- [About](https://usezn.com/about/) - company and product context
- [Contact](https://usezn.com/contact/) - support, privacy, legal, and security reports
- [Privacy](https://usezn.com/privacy/) - data handling and retention
- [CLI package](https://www.npmjs.com/package/zn-gate) - official npm package (`zn-gate`)
- [Pricing](https://usezn.com/pricing/) - plans and limits

## REST quickstart

Create a key at https://usezn.com/dashboard/ and call:

```http
POST https://api.usezn.com/prod/analyze
Authorization: Bearer <YOUR_API_KEY>
Content-Type: application/json

{"input":"text from an untrusted source"}
```

Block before tool execution when `verdict` is `block`. Successful and quota responses include `RateLimit-Limit`, `RateLimit-Remaining`, and `RateLimit-Reset`; 429 responses also include `Retry-After`.

## MCP

`POST https://api.usezn.com/mcp` uses Streamable HTTP and JSON-RPC. Authenticate with the same Bearer API key. The `analyze` tool accepts `{ "input": "string" }`.

## Local and CLI

The browser sandbox is at https://usezn.com/playground/. Install the CLI with `npm install -g zn-gate`; run `zn-gate init`, `zn-gate analyze`, or `zn-gate mcp`.