# Reviso documentation

How Reviso works for teams and AI agents: connect via MCP, explore the tool catalog, and get started with a specific client.

## MCP

Reviso exposes a Model Context Protocol (MCP) server over Streamable HTTP so AI agents can draft, review, and audit content with human approval.

- Endpoint: https://reviso.work/mcp

- Authentication: OAuth 2.1 with PKCE (browser flow) or a local account-agent token for headless agents.

- Authorized agents can only read and write content the connected account can already access.

- [Open the MCP setup guide](https://reviso.work/docs/mcp)

## Tools

The /docs/mcp-tools catalog lists every public MCP tool with its schema, read/write kind, and safety note — derived live from the server so it never goes stale.

- [Open the MCP tools catalog](https://reviso.work/docs/mcp-tools)

- [Quick tool summary](https://reviso.work/docs/mcp-tools-summary)

- [Agent workflow recipes](https://reviso.work/docs/mcp-recipes)

## Integrations

Step-by-step guides for connecting Reviso to ChatGPT, Claude, and Codex, including the connection URL, the OAuth prompt, expected permissions, and how to revoke access later.

- [Connect ChatGPT](https://reviso.work/integrations/chatgpt)

- [Connect Claude](https://reviso.work/integrations/claude)

- [Connect Codex](https://reviso.work/integrations/codex)

[Reviso documentation](https://reviso.work/docs)