glama-status-mcp

Daily-refreshed Glama TDQS score tracker for any MCP fleet. Scrapes per-tool docstring quality grades from glama.ai, stores in SQLite with snapshot history and delta tracking, surfaces via MCP tools and a 6-page web dashboard.

Features

• Auto-discover all your repos on Glama -- one command populates your fleet
• Per-tool 6-dimension breakdown -- Purpose, Usage, Behavior, Parameters, Conciseness, Completeness
• Snapshot delta tracking -- score changes between refreshes, stale repo detection
• LLM-powered analysis -- connected chat page and agentic tool (ctx.sample) to generate fixable todos
• Prefab UI cards for in-chat fleet overview and per-repo breakdown
• 6-page web dashboard -- Dashboard, Report, Tools, Chat, Help, Settings
• Tauri 2.0 NSIS installer -- single download, embedded backend, no Python required
• Track any author -- configure config/fleet-repos.json to monitor any Glama user's servers

Quick Install

Drag into Claude Desktop -- download glama-status-mcp-v0.1.1.mcpb and drop it onto the window. MCP tools only (no webapp).

For the web dashboard or native desktop app, see INSTALL.md.

What You Can Do

# Check fleet health
glama_status(operation="list")
glama_status(operation="report")

# Deep dive a repo
glama_status(operation="get", repo_name="email-mcp")
show_glama_repo_card(repo_name="blender-mcp")

# LLM-powered analysis
glama_agentic_analyze(repo_name="email-mcp")
glama_agentic_analyze()  # whole fleet

# Generate fix-todo reports
glama_generate_reports(repo_name="blender-mcp")
glama_generate_reports()  # all repos

# Auto-discover repos from Glama
glama_status(operation="discover")

Documentation

Doc	Contents
Installation	All install methods, prerequisites
Configuration	Env vars, fleet-repos.json, LLM setup
Glama Scoring Guide	TDQS explained: 6 dimensions, formula, grade thresholds, improvement workflow
Tool Reference	All 7 tools, 2 prompts, 12 operations
Development	Contributing, local setup, Tauri build

Requirements

• MCP client: Claude Desktop, Cursor, or any client supporting MCP tools
• For MCPB: Claude Desktop (drag-and-drop)
• For clone/run: Python 3.11+, uv
• For NSIS installer: Windows 10/11, WebView2 runtime (included in Win11)
• For LLM features: Ollama, LM Studio, or OpenAI API key

License

MIT