TS Score — Tool Activity Index

What is the TS Score?

The TS Score is our own 0–100 activity index for how “alive” a developer/AI tool is right now. It is not about stars or popularity — it measures whether a tool is actively maintained today. A popular project with no commits for six months scores low; a lesser-known one that ships weekly scores high. Use it to judge stability and longevity at adoption time.

Formula (full)

TS Score = recent commit (+30) + release (+20) + issue response rate (+20 × ratio) + downloads (up to +30), finally clamped with min(100).

Grades: A (80+) · B (60+) · C (40+) · D (below 40)

Component detail

Component Weight Measurement Source
Recent commit +30 Awarded if there is a commit within the last 30 days (last_commit_at) GitHub API
Release +20 Awarded if at least one release in the last 90 days GitHub Releases
Issue response rate +20 × ratio ratio = closed / (open + closed). round(20 × ratio) points GitHub Issues
Downloads up to +30 Larger of npm/PyPI 30-day downloads, mapped via the bracket table npm / PyPI

Download bracket table

30-day downloads (max of npm or PyPI) Points
1,000,000 or more +30
100,000 or more +24
10,000 or more +18
1,000 or more +12
100 or more +6
fewer than 100 +0

Recalculation example — LangChain (91)

Using data collected on 2026-06-16, LangChain’s TS Score recomputes as follows.

Component Observed Points
Recent commit last_commit_at = 2026-06-16 (within 30 days) +30
Release release within last 90 days +20
Downloads PyPI 30-day downloads ≥ 1,000,000 +30
Issue response rate ratio ≈ 0.55 → round(20 × 0.55) +11
Total min(100, 91) 91 → Grade A

30 + 20 + 30 + 11 = 91. Anyone can recompute the same value from the formula and observed inputs.

Why stars are excluded

GitHub stars reflect accumulated past popularity, not whether a project is maintained now. Repositories with tens of thousands of stars sit abandoned all the time. Because the TS Score answers “is it alive now,” stars are deliberately excluded from the formula. Star counts are shown separately on tool pages for reference only and never feed the score.

Open dataset

The source data behind the TS Score (tool_master) is published and downloadable as machine-readable CSV. The formula and figures on this page match the scoring code 100%.