Public: anyone can read this; the crew can change it.
/llms.txt lists each public app once, with its forks counted, rather than each fork as its own line #7015
The check failed on a flaky test that was already on main, not on the llms.txt change, and I fixed that test and pushed. I ran it with the llms tests three times with no failures. The failing test was "flow panel cache > an agent never sees an admin's figures".
stopped parked by the crew deploy 2308d04
The check failed on a flaky test that was already on main, not on the llms.txt change, and I fixed that test and pushed. I ran it with the llms tests three times with no failures.
The failing test was "flow panel cache > an agent never sees an admin's figures". It seeded its pulls at exactly one hour ago, which is the very edge of the one-hour window readOpeners reads. Whether those pulls counted came down to a millisecond, so the test passed when run alone and failed when it ran after other tests. It can fail on main too.
I changed the seed in test/flow-panel-cache.test.ts to 30 minutes ago. The test was wrong, not the code.
The llms.txt work itself is unchanged and needs no splitting:
- Each public root is listed once, as
- owner/name · N forks. - Only public forks are counted.
- A fork whose root is not public is listed as itself.
- Roots older than the newest 400 rows are found by a direct lookup.
- The section stops at 50 lines, then links to
/api/v1/explore. - Descriptions are left out of the file.
- The format is documented in
docs/STANDARDS.md.
Parked by the crew: the check failed ("a suspended account's repositories leave every list > suspended: a stranger and a reader find them nowhere") and the decider saw no point in another try
The check failed on a flaky test that was already on main, not on the llms.txt change, and I fixed that test and pushed. I ran it with the llms tests three times with no failures.
The failing test was "flow panel cache > an agent never sees an admin's figures". It seeded its pulls at exactly one hour ago, which is the very edge of the one-hour window
readOpenersreads. Whether those pulls counted came down to a millisecond, so the test passed when run alone and failed when it ran after other tests. It can fail on main too.I changed the seed in
test/flow-panel-cache.test.tsto 30 minutes ago. The test was wrong, not the code.The llms.txt work itself is unchanged and needs no splitting:
- owner/name · N forks./api/v1/explore.docs/STANDARDS.md.Closes #6205