tgfmcp
MCP server that enables Telegram bot interaction via Telegraf, providing tools for sending, replying, reacting, editing, deleting, forwarding messages, and receiving Telegram events over an optional notification channel.
README
tgfmcp
tgfmcp is an open-source Telegram stdio MCP server built on top of telegraf, commander, and @modelcontextprotocol/sdk.
It lets MCP-compatible clients interact with the Telegram Bot API through Telegraf and optionally subscribe to incoming Telegram events through an MCP notification channel.
Highlights
- Exposes Telegram as an MCP server over stdio.
- Uses
telegrafdirectly for polling and Bot API operations. - Supports interactive configuration via
tgfmcp configure. - Provides direct Bot API tools for bot identity and chat lookups.
- Includes mutating tools for sending, replying, reacting, editing, deleting, forwarding, and typing.
- Can emit incoming message events over an optional MCP notification channel.
- Stores downloaded incoming media attachments under
~/.tgfmcp/attachments/.
Requirements
- Node.js
24+ - A Telegram bot token
Installation
Use it without installing globally:
npx tgfmcp mcp
Or with Bun:
bunx tgfmcp mcp
If you prefer a global install:
npm install -g tgfmcp
For local development:
npm install
npm run build
npm run dev -- mcp
Quick Start
- Run the interactive configuration:
tgfmcp configure
This writes:
~/.tgfmcp/config.json
- Start the MCP server:
npx tgfmcp mcp
- If your MCP host supports notifications and you want incoming Telegram events, enable channels:
npx tgfmcp mcp --channels
The server uses stdio, so it is meant to be launched by an MCP client or wrapper rather than browsed directly in a terminal.
CLI Usage
MCP Server
npx tgfmcp mcp
bunx tgfmcp mcp
Starts the stdio MCP server for the configured Telegram bot.
Configure
npx tgfmcp configure
Then opens an interactive configure UI (Ink) to manage:
Bot tokenAllowed usersAllowed chats
Allowlist items are toggled from menu screens (select an entry to toggle it, then choose Back).
For enrolling users/chats, configure can generate a short code per screen and watch incoming Telegram updates. Send the code from the target user/chat and the matching ID is auto-added to allowlist.
Everything is persisted to:
~/.tgfmcp/config.json
MCP Tools
The server currently exposes these tools:
telegram_get_metelegram_get_statustelegram_get_chattelegram_get_chat_administratorstelegram_lookup_chattelegram_send_messagetelegram_send_media_from_base64telegram_send_media_from_pathtelegram_reply_to_messagetelegram_react_to_messagetelegram_edit_messagetelegram_delete_messagetelegram_forward_messagetelegram_send_typing
Push Channel
When started with --channels, the server:
- advertises the experimental MCP capability
hooman/channel - advertises
hooman/userwith pathmeta.user - advertises
hooman/sessionwith pathmeta.session - advertises
hooman/threadwith pathmeta.thread - advertises
hooman/channel/permissionfor remote daemon approvals - emits
notifications/hooman/channelfor incoming Telegram message events
If allowlist entries are configured, notifications/hooman/channel events are emitted only when either:
meta.session(chat ID) is inallowlist.chats, ormeta.user(sender user ID) is inallowlist.users
When no allowlist is configured (or both arrays are empty), all inbound channel events are emitted.
Each notification includes:
content: a JSON-encoded event payloadmeta.source: alwaystelegrammeta.user: the sender identity seed for the incoming messagemeta.session: the chat identity seed for the incoming messagemeta.thread: the Telegram message ID for the incoming message
The JSON-decoded content payload includes:
sourceselfmessagetext
If an incoming message contains downloadable media, the file is downloaded and included as a local attachment path in the event payload. Files are stored under ~/.tgfmcp/attachments/.
When Hooman sends notifications/hooman/channel/permission_request, tgfmcp posts the request back into the originating Telegram chat with inline approval buttons derived from params.options (defaults: allow once, always allow, deny). Button selections are relayed back over notifications/hooman/channel/permission.
Approvals are handled through inline buttons rendered from the permission request options.
Local Data
tgfmcp stores local state under ~/.tgfmcp/:
config.jsonfor bot token and allowlistattachments/for downloaded incoming media attachments
Notes
- Telegram bots cannot access arbitrary private chats; they can only interact where the bot has been added, contacted, or is otherwise permitted.
- Message mutation tools that target an existing message require explicit
chatIdandmessageIdinputs.
License
MIT. See LICENSE.
Recommended Servers
playwright-mcp
A Model Context Protocol server that enables LLMs to interact with web pages through structured accessibility snapshots without requiring vision models or screenshots.
Magic Component Platform (MCP)
An AI-powered tool that generates modern UI components from natural language descriptions, integrating with popular IDEs to streamline UI development workflow.
Audiense Insights MCP Server
Enables interaction with Audiense Insights accounts via the Model Context Protocol, facilitating the extraction and analysis of marketing insights and audience data including demographics, behavior, and influencer engagement.
VeyraX MCP
Single MCP tool to connect all your favorite tools: Gmail, Calendar and 40 more.
graphlit-mcp-server
The Model Context Protocol (MCP) Server enables integration between MCP clients and the Graphlit service. Ingest anything from Slack to Gmail to podcast feeds, in addition to web crawling, into a Graphlit project - and then retrieve relevant contents from the MCP client.
Kagi MCP Server
An MCP server that integrates Kagi search capabilities with Claude AI, enabling Claude to perform real-time web searches when answering questions that require up-to-date information.
E2B
Using MCP to run code via e2b.
Neon Database
MCP server for interacting with Neon Management API and databases
Exa Search
A Model Context Protocol (MCP) server lets AI assistants like Claude use the Exa AI Search API for web searches. This setup allows AI models to get real-time web information in a safe and controlled way.
Qdrant Server
This repository is an example of how to create a MCP server for Qdrant, a vector search engine.