mq-mcp

Local MCP server with OpenAI bridge and vector store Q&A for macOS.

macOS Python 3.11+ FastMCP OpenAI github.com/MCamner/mq-mcp →

What it does

Three tools that work together:

# Ask live questions using local MCP tools + OpenAI
bridget "List all Python files in the repo"
bridget "Show git diff and suggest a commit message"

# Search the vector store knowledge base
ask "How does bridge.py work?"
ask "Which tools are read-only?"

# Keyboard shortcut: Ctrl+O inserts "bridget " at the prompt

Architecture

bridget "prompt"
  └── bridge.py → OpenAI Chat API → MCP tool calls → server.py

ask "prompt"
  └── ask.py → OpenAI Responses API → vector store search

MCP Tools

read_repo_file

Read any file inside the repo root

list_repo_files

List files up to a chosen depth

search_repo

Full-text search with git grep

git_status

Branch, status, recent commits

git_diff

Current diff, repo or per file

get_system_resources

CPU, memory, disk usage

validate_project

Run scripts/validate.sh

run_mqlaunch

Open mqlaunch TUI in Terminal

update_repo_file write

Exact-match replace, no auto-commit

analyze_csv

Summarize a CSV file

analyze_guitar_pro

Parse GP3/GP4/GP5 files

open_in_app

Open a file in its default app

edit_image write

rotate, grayscale images

All tools are sandboxed to the repository root. Write-capable tools are clearly marked.

Install

# Clone and set up
git clone https://github.com/MCamner/mq-mcp
cd mq-mcp
./scripts/install.sh

# Check and run
mq-mcp doctor
mq-mcp tools
mq-mcp serve

Docs

Install guide · Security model · Demo examples · Integration map · Roadmap