The field notes
How we measure.
A star tells you someone bookmarked a repository. It won’t tell you whether the agent can sort your receipts, run on your laptop or keep your files private.
Metrics checked · How we measure
What makes the list?
You must be able to run the agent, give it a goal and have it choose and perform actions. A library, plugin or dashboard for someone else’s agent doesn’t qualify. The directory includes forks and independent alternatives; being listed does not imply an affiliation with OpenClaw.
The leaderboard orders available, non-archived projects by total GitHub stars. Ties are sorted by project slug. A star is an expression of interest, not a download, an active user or a recommendation. The catalog totals can include the last saved figures for historical projects; those projects don’t receive a current rank.
Two ways to count stars
- Stars added during a period
- GitHub’s stargazer history groups stars that are still present by the day they were added. If someone later removes a star, it won’t appear in that history. Daily charts use complete UTC days and preserve gaps; a missing observation is different from a measured zero.
- Net change over seven days
- We subtract the repository’s saved star count exactly seven calendar days earlier from its current count. This can be negative. Without that earlier snapshot there’s no net figure, even when the daily chart is available.
After the spike explains its own cohort and observation window. Its busiest day means the busiest day in that window.
What does “high issue load” mean?
For every 100 stars, how many open issues and pull requests does a repository have? We take GitHub’s open_issues_count and divide by stars. Low is below 1 per 100; moderate is 1 to under 5; high is 5 or more. With no stars, the ratio has no meaning.
A large queue may reflect an active community. Some maintainers track bugs elsewhere. The figure doesn’t tell you how quickly anyone will answer your question.
How the comparison score works
The health score is a rough repository activity measure out of 100. It is separate from issue load and doesn’t measure software quality, security or how well an agent completes a task.
- Open items relative to stars · up to 30 points
- Below 1% earns 30; 1% to under 3% earns 25; 3% to under 5% earns 15; otherwise 5. With no stars, this component earns no points.
- Last push · up to 30
- A push within one day earns 30; within seven days, 25; within 30 days, 15; older, 5. This checks GitHub’s last push timestamp, not the date a particular commit was written.
- Latest published release · up to 20
- Within seven days earns 20; within 30 days, 15; otherwise 5, including when GitHub lists no release.
- Seven-day net star change · up to 20
- A gain of at least 100 earns 20; 10–99 earns 15; 1–9 earns 10. Flat, falling or unknown growth earns no points. A missing baseline can therefore lower the score.
Recency is measured at the saved snapshot. Archived and unavailable repositories have no current score.
What have we actually checked?
Metrics come from GitHub’s public repository data, published releases and star history. A repository that cannot be reached retains its last successful snapshot and a separate last-checked date. We don’t fill the gap by guessing.
Setup and capability notes describe upstream documentation. A linked quick start is a place to begin, not a hands-on test or a security audit. If a source doesn’t establish a feature, we leave it unverified.
Corrections
Found a stale setup link, wrong capability or misleading chart? Send Dave Onkels the page and an upstream source.
· Clarified that the open-item count includes pull requests, that recency measures the last push, and that surviving star additions differ from net growth. Historical repositories no longer receive a current rank.