Scoring Methodology

Every MCP server gets a 0-100 quality score computed from 8 weighted parameters. Here's exactly how each parameter works:

Parameter Breakdown

Recency

When was the last commit? Active projects are more reliable.

20 points

Documentation

README length and quality. Longer, well-structured READMEs score higher.

15 points

Tests

Does the repo have test infrastructure? Jest, pytest, Mocha, or test files.

10 points

Auth/Security

README mentions auth patterns, API keys, tokens, OAuth, or credentials.

10 points

Star Velocity

Stars gained per month since creation. Growth signals community adoption.

15 points

Security Audit

README mentions security, validation, permissions. Penalized for high issue/star ratio.

15 points

Install Docs

Setup instructions count. Install, setup, usage, config, Docker, npx, pip keywords.

15 points

Activity Bonus

Recent commits in the last 30 days. Up to 5 bonus points.

5 bonus

Grade Mapping

Score 90-100 → A (Top Quality)
Score 70-89  → B (Good Quality)
Score 50-69  → C (Average)
Score 30-49  → D (Below Average)
Score 0-29   → F (Low Quality)

Data Sources

Scores are computed from the GitHub API. For each server we fetch:

Repository metadata (stars, forks, creation/push dates, language, license)
README content (length, keywords, install instructions)
File listing (test directories, package configs)
Recent commits (activity in last 30 days)

Servers without verifiable GitHub data are marked with a "⚠ no GitHub data" badge.