Compare public activity from each repository’s beginning, rather than giving both the same calendar window. The first 285 days are observed for both. The older repository’s additional history stays visible, but is not an equal-age comparison.
Activity, not a reliability ranking. Equal ages remove one distortion, not all of them. We do not know active installations, attempted upgrades, failure rates or operator hours. These charts cannot establish that one product is more dependable.
OpenClaw
Repository created 2025-11-24 Earliest retained non-prerelease GitHub release: 2025-11-25 Equal-age window ends 2026-09-04 (inclusive) Full observed span: 285 calendar days
Pi
Repository created 2025-08-09 Earliest retained non-prerelease GitHub release: 2025-12-02 Equal-age window ends 2026-05-20 (inclusive) Full observed span: 392 calendar days
Currently retained, published GitHub releases excluding drafts and prereleases. More releases do not mean more reliability.
All issue reports opened
Open and closed issues; pull requests excluded. Reporting volume is not the number of broken installations.
Reports currently labelled bug
Issues created in each window that carry the bug label now. Historical label states are not reconstructed.
Update / upgrade bug reports
Currently bug-labelled issues with update or upgrade in the title. This misses failures described only in the body.
Update / upgrade share of bug reports
First 285 days for both. Percentage of reports, not users or upgrades. Full 0–100% scale; differing labels and reporting practices still matter.
Audience now — not age-aligned
Current metadata snapshot at unequal repository ages. Historical star counts were not collected, so stars are not used to normalise the age-aligned reports. Compare star growth on Star History ↗. That live, third-party chart uses calendar dates, not aligned project ages, and is not part of this fixed snapshot.
The same first 285 days
Counts available at capture, grouped by creation or publication date. Links reproduce the issue queries; release links show the upstream release lists.
Counting from each repository’s earliest retained non-prerelease GitHub release instead gives a shared 277-day window. This excludes history before each first retained release, but still does not establish a public launch date. Earliest retained tags in this snapshot: OpenClaw: v0.1.1, Pi: v0.12.0. Neither starting point controls for audience or scope.
Starting-point sensitivity check, not the series plotted above. Source queries are included in the JSON snapshot.
Measure
OpenClaw
Pi
Starting date
2025-11-25
2025-12-02
End date (exclusive)
2026-08-29
2026-09-05
Non-prerelease releases
110
258
All issue reports opened
49,361
5,716
Reports currently labelled bug
13,832
1,654
Update / upgrade bug reports
619
68
Update / upgrade share of bug reports
4.48%
4.11%
What changed in the surrounding ecosystem?
The components matter as well as the core. These dates help locate the emerging alternatives, but repository creation is not proof that a particular feature was usable then.
Pi Livecraft and TelePi have their own issues, dependencies and maintenance work. They are not included in Pi’s core-repository counts. A fair comparison of complete setups must also measure those components and the work of assembling them.
How to read this fairly
“Inception” is an explicit proxy. Day zero is the UTC calendar date from GitHub’s created_at, not a verified first public launch or the coding agent’s birth. Pi’s repository covers a broader toolkit; repository creation and a tool’s public availability can differ. Renames, imported history and deleted records can affect both histories.
Windows are half-open. A point at day 30 includes the creation date through day 29. Activity on or after 2026-09-05 UTC is excluded. Samples are every 30 days plus the shared and latest endpoints; lines only connect those samples.
This is a retrospective count, not archived monthly snapshots. Open and closed issues are counted by creation date using labels and titles at capture. Releases are the retained, non-draft, non-prerelease GitHub records, dated by publication. Missing or deleted history is not reconstructed.
Age alignment trades away calendar alignment. Each first 285-day period took place in different months, with different models, audiences and ecosystem maturity. The guide’s same-calendar comparison answers a different, complementary question.
Reporting policies and scope differ. Pi documents auto-closing issues and PRs from new contributors by default; OpenClaw has a broader integrated surface. Labels, triage and user reporting habits can move these counts independently of reliability.
Stars are not exposure. Only current star counts were collected. They neither reconstruct historical popularity nor measure active users, so no reports-per-star reliability claim is made.
Do not substitute this for an operational benchmark. To test the personal experience, compare matched tasks, upgrade success, repair time, repair tokens, unwanted calls, permission scope and recovery across the whole installed stack.
All plotted counts and dates
Cumulative counts; end dates are exclusive. Day zero denotes an empty interval.