CodeView MCP Server

@mann-uofg

Basic information

About CodeView MCP Server

AI-powered code-review toolkit: MCP server + CLI to analyze GitHub PRs with local LLM smells, cloud LLM summaries, inline comments, risk gating, and test stub generation.

#codeview#code-review #ai-toolkit

CodeView MCP Server config

Add this server to your MCP-compatible client using the configuration below.

{
  "mcpServers": {
    "codeview-mcp": {
      "command": "python",
      "args": [
        "-m",
        "venv",
        ".venv",
        "&&",
        "source",
        ".venv/bin/activate"
      ]
    }
  }
}

CodeView MCP Server tools

Tool list not available yet

This listing does not currently have a tool list to display. This does not mean the server has no tools; check its documentation for supported capabilities.

We auto-extract tools from the README. The maintainer can list them under a ## Tools heading to populate this section.

Overview

What is CodeView MCP 🪄?

CodeView MCP 🪄 is an MCP server that performs a 30-second AI review of pull requests, combining static regex rules for critical smells, a local LLM (CodeLlama-13B) for quick heuristics, and a cloud LLM (Llama-3.1-8b-instant via Groq/OpenAI) for human-style summaries and risk scores. It targets developers who need fast, privacy-conscious PR analysis.

How to use CodeView MCP 🪄?

Clone the repository, create a Python virtual environment, install dependencies, and install the package with pip install -e .. Run a smoke test with reviewgenie/codeview ping https://github.com/psf/requests/pull/6883. Store secrets such as a GitHub PAT and Groq/OpenAI API key using the codeview_mcp.secret keyring module.

Key features of CodeView MCP 🪄

Static regex rules for critical code smells
Local LLM heuristics with no cloud cost
Cloud LLM human-style summary and risk score (0–1)
One-click inline comment accept or ignore
SQLite diff cache and ChromaDB hunk embeddings
OpenTelemetry tracing and GitHub back-off logic

Use cases of CodeView MCP 🪄

Reviewing large PRs quickly for security or performance issues
Running a CI gate that blocks merges when risk score exceeds a threshold
Generating stub test files and opening test PRs automatically
Auditing code with inline comments that can be accepted or dismissed

Frequently asked questions

What does the `analyze` tool produce?

It returns a summary, a list of smells, rule hits, and a risk score between 0 and 1, with typical latency of 6–10 seconds.

How does CodeView MCP 🪄 protect code privacy?

Only the diff snippet is sent to the cloud LLM (Groq); the full codebase never leaves your machine.

What LLMs does CodeView MCP 🪄 use?

It uses CodeLlama-13B locally and Llama-3.1-8b-instant via a cloud provider (Groq or OpenAI, configured via `OPENAI_API_KEY` and `OPENAI_BASE_URL`).

What are the runtime dependencies?

Python 3.10+, SQLite, ChromaDB, and network access for cloud LLM calls. Secrets are stored via the system keyring.

How is the server invoked or transported?

The README shows a CLI command (`reviewgenie/codeview`) and smoke test; the MCP transport is not specified further.

Comments

More Developer Tools MCP servers

FARPY

mangomunchr

Verified distributed compute for humans, applications, and agents. Submit Blender, Octane, and video workloads, route them across qualified GPUs, track asynchronous execution, download finished artifacts, and verify FARP

9apes

The 9apes MCP server lets an AI assistant read the GitHub Actions data 9apes already collects for your organization: runs, jobs, steps, and their durations. Ask it why a build got slower, which jobs take the longest, or

GripForge

gripforgeai

The tools your AI needs to make your games. Turn prompts into production-ready game assets — animated characters with their weapons attached, seamless textures, terrain, VFX, HUDs — and playable game kits Unity, Godot, U

Vibgrate MCP

Vibgrate

Query your team's drift, vulnerability, and migration data from any AI assistant. Vibgrate MCP connects Cursor, Claude, ChatGPT, Windsurf, or VS Code to Vibgrate Cloud: 51 tools for DriftScores, CVEs and EOL runtimes, up

Reelier

Maxime Houle

Agents make claims. Reelier writes receipts — record an agent's tool-call workflow once, replay it deterministically at 0 tokens, and diff runs to catch drift.