Ecosystem Activity
A snapshot of community contribution patterns across the ecosystem
Learn more about how it works.
Organic 67.0% -15.9%
Mixed 10.9% +12.6%
Automation 22.1% +42.2%
A snapshot of community contribution patterns across the ecosystem
Learn more about how it works.
Organic 67.0% -15.9%
Mixed 10.9% +12.6%
Automation 22.1% +42.2%
This page tracks the state of the GitHub community by analyzing public activity. The question we're trying to answer: how many accounts are genuinely contributing to open source, and how many are just flooding it with noise?
This isn't an anti-AI project, though we do have our opinions. We're simply tired of watching open source maintainers drown in spam.
Every hour, we check each of our repositories and collect 10 of the most recent pull request opened during the hour before, or however many there are.
We skip accounts that match known bots, like Copilot, Dependabot, Renovate, GitHub Actions, and other common automation tools. For every PR author, we review up to 300 of their recent public GitHub events, using the same identity library that powers every AgentScan profile scan.
The hourly view shows the last 25 hours, one point per scan. Once a day is complete, we compound its hours into a single daily scan result, which is what feeds the daily view and its 2 months of history.
These are indicators, not definitive verdicts. The chart describes a sample of pull requests from a fixed list of repositories, so read it as a trend over time rather than as a measurement of GitHub as a whole.
For now, we track 200 repositories. Some we picked by hand because we care about them; others were chosen at random from the GitHub trending repositories page. They go from TypeScript, JavaScript, Go, Python, and Rust to toolchains, frameworks, testing libraries, learning websites, and even AI-related projects.
The list keeps growing. We would love to cover many more projects, but we have to stay within GitHub's API rate limits.