1. Conduid
  2. Search
  3. Fastmcp Pr Review
MCP server · Search

Fastmcp Pr Review

FastMCP PR review server — three examples of increasing complexity (structured output → tool calling → production pipeline)

Unclaimed last commit 6 months ago search
34Low

Scored 6 months ago · breakdown

About Fastmcp Pr Review

Fastmcp Pr Review is an MCP server published by strawgate in the Search category: fastMCP PR review server — three examples of increasing complexity (structured output → tool calling → production pipeline).

Six months or more without a commit doesn't mean the server is broken, but check the open issues before depending on it in production.

Install

From source
git clone https://github.com/strawgate/fastmcp-pr-review

This server has no ConduID identity, so agent calls to it are not receipted. Pin the version you install and review the source before granting it credentials.

Tools (2)

review_pr_fast`fast.py`

Parameters not published.

review_pr_thorough`thorough.py`

Parameters not published.

Ask AI

✦

Ask AI about Fastmcp Pr Review

Powered by Claude · Grounded in docs

I know everything about Fastmcp Pr Review. Ask me about installation, configuration, usage, or troubleshooting.

Security checks

  • ✓README presentThe repository ships documentation.

5 checks have not been run yet.

README

FastMCP PR Review

A GitHub PR review server built with FastMCP, with two review modes: fast and thorough.

Each mode is self-contained — read either file top-to-bottom to understand the complete pattern.

Review Modes

Tool File What it does
review_pr_fast fast.py One LLM call, structured output, no tools
review_pr_thorough thorough.py Multi-pass pipeline: filter → review → verify

Fast

One ctx.sample() call with result_type=PRReviewResult. Sends the full diff, gets back a structured review. Best for small PRs or quick checks.

Thorough

Multi-pass pipeline with batched file filtering, per-file review with a verification protocol, and agentic verification where the LLM explores the repo to confirm or disprove findings. Configurable intensity (conservative/balanced/aggressive). Only confirmed findings survive.

Quick Start

uv sync
cp .env.example .env
# edit .env with your GitHub and Gemini credentials

uv run fastmcp-pr-review
# or override runtime settings explicitly
uv run fastmcp-pr-review --gemini-model gemini-2.5-pro
make server-stdio
make server-http HOST=0.0.0.0 PORT=8000

The server auto-loads .env when started from the repo directory, and the Click launcher also reads GITHUB_TOKEN and GEMINI_API_KEY directly from the process environment. That means hosted environments like Horizon can inject those vars without extra CLI flags.

The server also exposes data tools: get_pr_info, get_pr_diff, get_pr_files.

Deploy to Horizon

For Horizon-style deployments that import a module and look for an mcp object, point it at the repo-root run_horizon_server.py. That file loads env vars and exposes:

mcp = create_server()

Set GITHUB_TOKEN and GEMINI_API_KEY in Horizon, and it can import run_horizon_server.py directly.

Add to Claude

Add the server to your Claude MCP config with a stdio entry that runs this repo through uv:

{
  "mcpServers": {
    "fastmcp-pr-review": {
      "type": "stdio",
      "command": "uv",
      "args": [
        "--directory",
        "/absolute/path/to/fastmcp-pr-review",
        "run",
        "fastmcp-pr-review",
        "--gemini-model",
        "gemini-2.5-pro"
      ],
      "env": {}
    }
  }
}

Keep GITHUB_TOKEN and GEMINI_API_KEY in /absolute/path/to/fastmcp-pr-review/.env or inject them via the host environment so Claude does not need inline secrets in its MCP config. Use CLI flags for non-secret runtime settings like the fallback Gemini model.

Makefile

The repo includes a Makefile for the common local workflows:

make server-stdio
make server-http HOST=0.0.0.0 PORT=8000
make format
make lint
make lint-fix
make typecheck
make test
make build

Development

See DEVELOPING.md for testing and sampling patterns, and docs/ARCHITECTURE.md for architecture and design decisions.

License

MIT

README copy captured when this server was indexed; it may lag the repository. The original is authoritative.

Questions

About Fastmcp Pr Review

How do I install Fastmcp Pr Review?

No package has been published for this server, so install it from source: clone the repository and follow its README.

Is Fastmcp Pr Review safe to use with an AI agent?

Its trust score is 34 out of 100 (low). It passes 1 of the 1 static security checks run so far. It has no ConduID identity yet, so agent calls to it are not receipted.

Is Fastmcp Pr Review still maintained?

The last commit was 6 months ago. That's long enough that you should check whether the maintainer is responding to issues before depending on it.