Click on "Deploy Server".
Wait a few minutes for the server to deploy. Once ready, it will show a "Started" state.
In the chat, type @ followed by the MCP server name and your instructions, e.g., "@SYNTARO MCP Label issue #42 as syntaro:fix and open a PR when the fix is ready"
That's it! The server will respond to your query, and you can continue using it as needed.
Here is a step-by-step guide with screenshots.
Open-source MCP server + agent skills for SYNTARO: an AI bot that investigates a labeled GitHub issue, writes a fix with regression tests, and opens a pull request. Any MCP-capable agent (OpenCode, Claude Code, Cursor, Codex) can drive it.
Hosted endpoint (target, pending deploy): https://mcp.syntaro.io/sse (SSE) and
https://mcp.syntaro.io/mcp (streamable HTTP) — DNS is not live yet, so run
it yourself below or deploy with docs/HOSTING.md, then update
server.json transports to the live URLs.
npm: @aimino/syntaro-mcp (see docs/PUBLISHING.md)
Skills: skills/ — syntaro (submit→poll loop) plus
cavecrew / caveman / ponytail (plan-first + verification-tail
discipline) and automation-testing.
License: AGPL-3.0-only (LICENSE).
pip install -r requirements.txt
python -m syntaro_mcp.server stdioOr via npm (runs the same Python server under the hood):
npm install -g @aimino/syntaro-mcp
npx syntaro-mcp stdioRegister with your agent — bash install.sh --opencode (also --claude,
--cursor, --codex), or point a remote client at the hosted endpoint:
{ "mcpServers": { "syntaro": { "url": "http://localhost:4095/sse" } } }Remote (after deploy): replace the URL with https://mcp.syntaro.io/sse.
Full client matrix: docs/INSTALL.md.
Related MCP server: GitPilot MCP
Tool | What it does |
| Label a GitHub issue ( |
| Trigger the fix pipeline for an issue URL → |
| Poll a fix run by |
| PR URL + details for a completed run |
| Tracked issues with fix status |
| Symbol/file/pattern search |
| Linear ticket check + create |
| Persistent agent memory files |
| Post to a Slack channel/thread |
| Rehydrate a conversation's |
Resources: syntaro://runs/{run_id}, syntaro://issues/{issue_id}.
Variable | Purpose | Default |
| Syntaro backend for fix runs |
|
| Backend auth (remote SSE clients send it as a header) | — |
| GitHub auth for labeling | — |
| Enables | — (tools report "not configured") |
| Enables | — |
|
|
|
| SSE port |
|
| Optional shared secret for the hosted endpoint | — |
No secrets ship in this repo — everything arrives via env at runtime.
docker build -t syntaro-mcp .
docker run -p 4095:4095 -e SYNTARO_API_KEY=... syntaro-mcp
curl localhost:4095/health # {"status":"ok","service":"syntaro-mcp"}Cloud Run + custom domain: docs/HOSTING.md. Marketplace releases (npm, MCP registry, Smithery, Glama): docs/PUBLISHING.md.
syntaro_mcp/ Python MCP server (FastMCP): server, handlers, agent_handlers
workers/ Pipeline client (hosted API first, embedded engine fallback)
skills/ Agent skills (syntaro + workflow skills)
.claude-plugin/ Claude marketplace manifest
server.json MCP registry manifest
smithery.yaml Smithery deploy config (+ Dockerfile.smithery)
index.js npm wrapper (spawns the Python server)
deploy/ Cloud Run service + deploy script
docs/ HOSTING / PUBLISHING / INSTALL / Glama listingSee CONTRIBUTING.md and SECURITY.md. The commercial platform (webhook server, billing, dashboard) lives in the private monorepo; this repo is the full open-source distribution surface.
This server cannot be deployed
The AI orchestration agent for modern software teams.
Git-backed platform for skills, tools, and context for AI agents
Connect AI assistants to GitHub - manage repos, issues, PRs, and workflows through natural language.
Code intelligence platform for AI agents. 20 tools for architecture, security & impact analysis.
