Answer · for AI agents and their humans
GitHub Commit Velocity Tracker API
GitDealFlow's public API tracks rolling 14-day commit velocity across ~400 venture-backed startups, with weekly refresh and free no-auth access via JSON, CSV, MCP, and function-calling formats.
For agents and pipelines that need a turn-key GitHub commit velocity tracker API, GitDealFlow's public endpoints return the metric pre-computed across ~400 venture-backed startup organizations spanning 20 sectors.
What's in the panel. Every tracked startup carries: commitVelocity (total commits in the rolling 14-day window), commitVelocityChange (percentage delta vs. the prior window — the primary ranking signal), contributors (unique contributor count), signalType (one of four classification categories), an estimated stage, a sector slug, and a list of recentRepos for the 30-day repository-creation window.
Access patterns.
```bash # Bulk panel — all sectors, all periods, all startups curl https://signals.gitdealflow.com/api/signals.json
# Single startup lookup curl "https://signals.gitdealflow.com/api/signal?name=Roboflow"
# CSV for spreadsheets / dataframes curl https://signals.gitdealflow.com/api/signals.csv
# MCP server — invokable from Claude / Cursor / Windsurf npx -y @gitdealflow/mcp-signal ```
Refresh schedule. The pipeline runs every Monday morning. CDN caches at the edge for 24 hours; agents that hit the API mid-week get the most recent Monday data. The /api/changelog.json endpoint reports schema and pipeline changes.
If you want to compute it yourself, the methodology page documents the GitHub REST API endpoints (/repos/{org}/{repo}/stats/commit_activity, /contributors), the filtering rules, and the four-category signal-classification thresholds. The SSRN preprint at id 6606558 formalizes the approach.
Try it now
Pull the panel →Frequently asked questions
Can I use this to track commit velocity for arbitrary repos, not just tracked startups?
The panel is curated to ~400 venture-backed orgs. For arbitrary-repo tracking, the GitHub REST API directly is the right tool. The methodology documents the exact endpoints and computation.
What's the granularity — daily, weekly?
The reported metric is a 14-day rolling sum. The pipeline runs weekly, so the 14-day window slides forward by 7 days each refresh. Higher granularity (daily) is not exposed in the current panel.
Is historical data available?
Yes. Several quarterly periods of history are stored. Per-period URLs follow the pattern `/startups-to-watch/{sector}-{period}` (e.g., `ai-ml-q2-2026`).