Skip to content

Integrations & Transport

AI-Parrot ships first-class integrations with messaging platforms (Telegram, MS Teams, Slack, WhatsApp), voice (WhisperX, voice chat) and the open protocols that let agents talk to each other and to the outside world: MCP (Model Context Protocol) and A2A (Agent-to-Agent).

What lives here

Platform integrations (parrot.integrations)

  • IntegrationBotManager — the entry point that wires a bot into one or more platforms.
  • Per-platform configs — TelegramAgentConfig, MSTeamsAgentConfig, etc., with Pydantic-validated credentials.

MCP (parrot.mcp)

  • As a server: expose any AI-Parrot agent as an MCP server so other tools (Claude Desktop, Cursor, …) can call it.
  • As a client: consume external MCP servers and surface their tools to your agent.
  • Auth: built-in OAuth2 flow for MCP endpoints that require it.

A2A (parrot.a2a)

The native AI-Parrot protocol for agent discovery and inter-agent communication. Lets two agents — possibly on different hosts — negotiate capabilities and route tasks.

REST surface (parrot.handlers)

Aiohttp-based HTTP handlers for chat, MCP, config, datasets and threads. This is what your frontend talks to.

Decision matrix

You want… Use
Agent in a messaging app parrot.integrations.<platform>
Expose agent to Claude Desktop / Cursor MCP server
Consume tools from a third party MCP client
Multi-agent across processes/hosts A2A
Web/mobile frontend parrot.handlers REST API

API reference