$ timeahead.in

$ cat about.md

About MCP Score

An independent registry that scores every public MCP server on security, freshness, adoption, quality, and trust — updated nightly, no vendor money, no ads.

“With great power comes great responsibility.”

— Uncle Ben, Spider-Man (2002)  ·  applies equally to MCP servers with exec permissions

// why this exists

The MCP ecosystem is growing fast. Hundreds of servers are published every month — most of them open-source side projects with no security review, no maintenance commitment, and no way for end-users to tell the good ones from the risky ones.

MCP Score automates what you'd want a senior engineer to do before connecting an unknown server to your AI agent: scan for leaked credentials, check if it's still maintained, verify the install works, and summarise the risk in a single number you can act on.

// scoring methodology

Every server is scored 0–100 across five dimensions. Scores are recomputed nightly. The formula is public and versioned — no secret sauce, no pay-to-rank.

Security35%Gitleaks secret scanning, dependency vulnerabilities, presence of known malicious patterns.

Freshness25%Days since last commit, last npm/pypi publish, release cadence over the past year.

Adoption20%Weekly downloads, GitHub stars, fork count — signals of real-world community usage.

Quality10%README length, license presence, CI setup, test coverage indicators.

Trust10%Verified publisher claim, signed releases, install success rate in our nightly sandbox.

// independence & transparency

✗ No vendor money

We accept no payment from MCP server authors. A score cannot be bought.

✗ No ads

No ad networks, no sponsored placements, no affiliate links.

✓ Open methodology

The scoring algorithm is documented, versioned, and open to public dispute.

✓ Nightly updates

All scores are recomputed every night from fresh crawl data.

// built by

Nikhil Patni

Independent developer. Built MCP Score because I wanted a trustworthy way to evaluate servers before connecting them to my own AI agents — and nothing like it existed.

⌥ github.com/nkpatnis ⌥ linkedin.com/in/nkpatnis

// content policy

Server metadata (names, descriptions, download counts, stars) is fetched from public registries (npm, PyPI, GitHub) under their respective Terms of Service. Titles and excerpts are reproduced for indexing and display purposes only — full credit and copyright remain with the original authors and publishers.

If you believe a score is incorrect, use the dispute form on the server's detail page. We review every dispute within 48 hours.