openpoem/spec-score-mcp

Score your specs before feeding them to an LLM. MCP server that measures spec quality on 4 axes.

43
/ 100
Emerging

Implements Claude-based scoring on four dimensions (completeness, clarity, constraints, specificity) with a balance metric that prioritizes even coverage over individual axis strength, exposed as three MCP tools: `spec_score`, `spec_visualize`, and `spec_compare`. Integrates with Claude Code via slash commands and MCP protocol, plus available as an HTTP API powering a ChatGPT Custom GPT and standalone web interface. Uses normalized vector math to measure axis variance, with verdicts (SHIP IT, ALMOST, DRAFT, etc.) determined by balance thresholds and pattern matching rather than raw scoring.

Available on npm.

Maintenance 10 / 25
Adoption 2 / 25
Maturity 18 / 25
Community 13 / 25

How are scores calculated?

Stars

2

Forks

2

Language

TypeScript

License

MIT

Last pushed

Feb 28, 2026

Commits (30d)

0

Dependencies

1

Get this data via API

curl "https://pt-edge.onrender.com/api/v1/quality/mcp/openpoem/spec-score-mcp"

Open to everyone — 100 requests/day, no key needed. Get a free key for 1,000/day.