notify-mcp
Enables AI to send push notifications to your phone via a configurable gateway. It is a thin wrapper that posts to an Apprise push gateway.
README
messagers — MCP push-notification tool
English | 简体中文
An MCP server that gives any MCP client (Claude Code, etc.) a send_notification
tool. The AI calls it to proactively push a short message to your phone — no need
to read any doc or remember a curl/token. It is a thin wrapper: the tool just
POSTs to the Apprise push gateway; all chat credentials / proxy / channels live
in the gateway.
Claude / MCP client ── send_notification(message, title?, tag?) ──► notify_mcp.py
│ HTTP POST (+bearer)
▼
apprise gateway ──► Telegram / Feishu / ...
Point the wrapper at your gateway with NOTIFY_GATEWAY_URL (see Configure). The
gateway itself (server + how to deploy it) is documented in
docs/NOTIFY_GATEWAY.md; that file describes one example
deployment (Linux + systemd) — host, port and proxy will differ per machine.
The tool
send_notification(message: str, title: str = "", tag: str = "default") -> str
— pushes message (with optional bold title) to the recipient group tag.
Returns "ok" or an error string.
Install
Dependencies are managed with uv (Python 3.12,
pinned in .python-version; lockfile uv.lock is committed):
uv sync # wrapper deps (mcp) + dev tools (pytest, pre-commit)
uv sync --extra gateway # also install apprise, for running the gateway locally
Scripts
| Script | What it does |
|---|---|
scripts/install_mcp.sh |
Register this MCP with whichever clients are installed (Claude Code user scope + Codex CLI global). Idempotent. |
scripts/start_gateway.sh |
Run the Apprise push gateway (server.py) in the foreground, with sensible HOST/PORT defaults. |
./scripts/install_mcp.sh # add the MCP to Claude Code + Codex CLI
./scripts/start_gateway.sh # start the gateway (needs ./token and ./targets.json)
A real deployment usually runs the gateway under a service manager (e.g. systemd) —
see docs/NOTIFY_GATEWAY.md. start_gateway.sh is for
local/manual runs.
Configure
Config is via environment variables (the tool reads them at runtime):
| Env | Default | Meaning |
|---|---|---|
NOTIFY_GATEWAY_URL |
http://localhost:27319/notify |
gateway endpoint — set this to your gateway's host:port |
NOTIFY_GATEWAY_TOKEN |
— | bearer token (overrides the file) |
NOTIFY_TOKEN_FILE |
./.token next to the script |
file holding the bearer token |
The default assumes the gateway runs on the same host. On any other machine, set
NOTIFY_GATEWAY_URL to where the gateway actually listens (e.g. when registering,
./scripts/install_mcp.sh bakes in NOTIFY_GATEWAY_URL if you export it first).
Put the gateway bearer token in a gitignored .token file (recommended — keeps
the secret out of both git and the MCP client config):
printf '%s' '<BEARER_TOKEN>' > .token && chmod 600 .token
Register with Claude Code / Codex CLI
Easiest — from the repo directory, run the install script (it auto-detects the
repo path and registers with both clients using a uv run launch command, so the
server uses this repo's locked deps):
# optionally point at your gateway first; it gets baked into the registration:
export NOTIFY_GATEWAY_URL=http://<gateway-host>:<port>/notify
./scripts/install_mcp.sh
Or register manually (replace <repo> with this repo's absolute path):
# Claude Code, user scope (available in every project):
claude mcp add notify -s user -- uv run --directory <repo> python notify_mcp.py
# Codex CLI (writes to ~/.codex/config.toml; verify with `codex mcp list`):
codex mcp add notify -- uv run --directory <repo> python notify_mcp.py
Restart / reconnect the client so it picks up the new server. The
send_notification tool then appears automatically — the AI reads the tool
description, not this README.
To point at a gateway, add -e NOTIFY_GATEWAY_URL=http://<gateway-host>:<port>/notify
(Claude) / --env ... (Codex).
Windows: the
.shscripts need Git Bash / WSL. Without them, register manually as above (usewhere uvto confirmuvis on PATH) and use a native path for<repo>.
Test (without an MCP client)
uv run python -c "from notify_mcp import _post; print(_post('test from notify-mcp', title='hello'))"
# -> (True, '{"ok": true, ...}') and a chat message arrives
Unit tests
uv run pytest -q # 28 tests, no network needed
tests/ covers the wrapper (test_notify_mcp.py), the gateway with a stubbed
apprise (test_server.py), and the MCP introspection contract
(test_introspection.py). None of them touch the real network or gateway.
Lint / format (pre-commit)
Formatting and linting are enforced with ruff
via pre-commit; uv-lock keeps uv.lock in sync with pyproject.toml.
uv run pre-commit install # once per clone — runs the hooks on every commit
uv run pre-commit run --all-files
Docs
All under docs/:
docs/NOTIFY_GATEWAY.md— example gateway deployment (systemd, proxy, Telegram/Feishu).docs/MCP_INTROSPECTION.md— how a client discovers this server's identity and tool (the tool description IS thesend_notificationdocstring; there is no URL path).docs/ABOUT.txt— short, human-friendly intro (中文).
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.