Read and send Telegram messages through your assistant — chats, channels and history via the client API.
Telegram's MCP servers take an unusual route: instead of a bot API (which can't read arbitrary chats), the leading implementations use Telegram's client API via MTProto — the assistant effectively acts as your own Telegram account, seeing what you see. That makes them far more capable than a typical bot bridge, and correspondingly more sensitive.
With it configured, your assistant can list your dialogs, read history from chats, groups and channels you're in, and send messages as you. The value concentrates in triage and digestion: Telegram power users accumulate dozens of channels and group chats, and "summarise what mattered across my feeds this morning" is exactly the kind of task an assistant does better than a thumb-scrolling human. Extracting action items from a chaotic group planning thread, catching up a muted channel, drafting replies in your voice for approval — all natural fits.
Setup reflects the client-API approach: you register an application at my.telegram.org to get an api_id and api_hash, then complete a one-time login (phone code, plus 2FA password if set) to produce a session the server reuses. Implementations vary — chigwell/telegram-mcp (Python, Telethon) is a full-featured choice with dozens of tools; chaindead's Go implementation focuses on a tighter dialogs/read/send core.
Be clear-eyed about the trust model: the session string is a key to your entire Telegram account, so store it like a password, and let the client's tool-confirmation flow gate anything that sends. Used with that care, it turns Telegram from an attention sink into something you consult on your terms.
Dozens of channels condensed into one 'what actually mattered' briefing.
Action items and decisions extracted from sprawling planning threads.
Responses drafted in your voice, sent only after you approve.
| Tool | What it does |
|---|---|
| get_chats | List your dialogs (chats, groups, channels) with pagination |
| list_messages | Read message history from a chat with filters |
| send_message | Send a message to a chat as your account |
| get_contacts | List contacts |
| search_public_chats | Search public channels and groups |
| get_me | Your own account info |
{
"mcpServers": {
"telegram": {
"command": "uv",
"args": ["--directory", "/path/to/telegram-mcp", "run", "main.py"],
"env": {
"TELEGRAM_API_ID": "your-api-id",
"TELEGRAM_API_HASH": "your-api-hash",
"TELEGRAM_SESSION_STRING": "your-session-string"
}
}
}
}Get api_id/api_hash from my.telegram.org; generate the session string once with the repo's session_string_generator.py.
Telegram API credentials from my.telegram.org and a one-time login to generate a session string. Python with uv.
| Variable | Description | Required |
|---|---|---|
| TELEGRAM_API_ID | Application ID from my.telegram.org | Yes |
| TELEGRAM_API_HASH | Application hash from my.telegram.org | Yes |
| TELEGRAM_SESSION_STRING | Login session produced by the one-time auth flow — treat as a full account credential | Yes |
Your assistant inside the workspace — read channels, search history, post messages and tame the noise.
Inbox intelligence — search, read, draft and send Gmail through your assistant with OAuth auto-setup.
Read and write Jira, Confluence, Bitbucket, JSM and Compass from your AI client — with your own permissions.
MCP server enabling room-based messaging between multiple agents
AI-powered access to Attio CRM. Manage contacts, companies, deals, tasks, and notes. Search records, update pipelines, and automate workflows for
Real availability, real bookings, and single-use scheduling links — driven from the conversation.