telegra-me
Bridges MCP clients to Telegram for sending notifications and waiting for user replies.
README
telegra-me
MCP server that connects AI agents to your private Telegram bot — send notifications and ask questions while you are away from your computer.
Website
Website: https://guedes-joaofelipe.github.io/telegra-me/
Static product landing (website/), deployed to GitHub Pages on push to main. The site is available after the first successful workflow run and enabling Pages (source: GitHub Actions) in repository settings.
Problem
Long-running AI agents may need your input while you are away from your desk. telegra-me bridges MCP clients on your computer to Telegram on your phone so agents can notify you and wait for replies.
Prerequisites
- uv — install dependencies with
uv syncafter cloning - A Telegram bot and chat ID — follow docs/SETUP.md
- Configuration reference: docs/CONFIG.md
Quick start
git clone <repository-url>
cd telegra-me
uv sync
cp config.example.yml config.yml
# Edit config.yml with your Telegram token, chat ID, and API key
make serve
The daemon listens on http://127.0.0.1:8765/mcp by default.
Background service
Install a user service that starts at login (no terminal required):
uv sync
make service-install
| Command | Purpose |
|---|---|
make service-install |
Install and start launchd (macOS) or systemd user service (Linux) |
make service-uninstall |
Stop and remove the service |
make service-status |
Print whether the daemon is running |
Logs on macOS: ~/Library/Logs/telegra-me/stdout.log and stderr.log.
On Linux, enable lingering if the service must run without an active session:
loginctl enable-linger "$USER"
MCP clients
telegra-me is not Cursor-specific. The daemon speaks standard MCP over HTTP on loopback with Bearer API key auth. Start it first with make serve or make service-install.
Cursor
~/.cursor/mcp.json or project .cursor/mcp.json:
{
"mcpServers": {
"telegra-me": {
"url": "http://127.0.0.1:8765/mcp",
"headers": {
"Authorization": "Bearer YOUR_MCP_API_KEY"
}
}
}
}
Claude Code (CLI)
claude mcp add --transport http telegra-me http://127.0.0.1:8765/mcp \
--header "Authorization: Bearer YOUR_MCP_API_KEY"
Claude Desktop
Local Claude Desktop config is stdio-only. Use mcp-remote to bridge stdio → the running HTTP daemon (~/Library/Application Support/Claude/claude_desktop_config.json on macOS):
{
"mcpServers": {
"telegra-me": {
"command": "npx",
"args": [
"-y",
"mcp-remote",
"http://127.0.0.1:8765/mcp",
"--transport",
"http-only",
"--header",
"Authorization:${MCP_API_KEY_HEADER}"
],
"env": {
"MCP_API_KEY_HEADER": "Bearer YOUR_MCP_API_KEY"
}
}
}
}
Replace YOUR_MCP_API_KEY with server.api_key from config.yml or the MCP_API_KEY environment variable.
Note: Claude Custom Connectors (cloud-brokered remote MCP) require a public HTTPS URL and do not work with a localhost-only daemon.
MCP tools
| Tool | Purpose |
|---|---|
send_telegram_message |
One-way text notification |
send_telegram_photo |
Send an image (HTTPS URL or local absolute path) |
send_telegram_audio |
Send an audio track with optional title/performer |
send_telegram_voice |
Send a voice note (OGG/OPUS recommended) |
ask_telegram_user |
Ask a question and wait for a reply (text, photo, audio, or voice) |
Media source rules:
- Provide exactly one of
*_url(HTTPS only) or*_path(absolute local file path) per call ask_telegram_useraccepts optionalphoto_url/photo_pathto attach an image; the question becomes the caption- Answered
ask_telegram_userresponses includereply_type(text,photo,audio, orvoice); media replies include Telegramfile_id(bytes are not downloaded back to the agent)
Voice notes work best as OGG files with OPUS encoding (Telegram’s native voice format).
Troubleshooting
| Symptom | Check |
|---|---|
| Daemon exits immediately | config.yml missing telegram.bot_token, server.api_key, or empty allowed_chat_ids — see docs/CONFIG.md |
| MCP client cannot connect | Daemon running? (make service-status or make serve) |
| HTTP 401 from MCP client | Authorization: Bearer <api_key> must match server.api_key or MCP_API_KEY |
| Telegram messages not received | Bot token valid? chat_id in allowed_chat_ids? Did you /start the bot? |
ask_telegram_user times out |
Reply with a text, photo, audio, or voice message from the allowlisted chat while the tool is waiting |
| Media send fails after upgrade | Restart the background service (make service-uninstall && make service-install, or launchctl kickstart -k gui/$UID/com.telegra-me.mcp on macOS) so launchd loads the new code |
| Linux service stops at logout | Run loginctl enable-linger "$USER" |
Development
make lint
make tests
make coverage
Feature specs: docs/specs/features/core-platform/, docs/specs/features/media-messaging/.
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.