ouroboros/devtools
Ouroboros 9f90206c25 CI: run no workflow on a schedule; the keyless E2E runs on dispatch and tags
Nobody read the nightly results (owner, 2026-10-05), and the nightly
keyless E2E could not have caught a development regression anyway: a
scheduled run checks out the default branch, which is the last release,
so the acceptance-scenario hang on ouroboros (#1461) stayed invisible
for five days. The schedule block is gone; system-e2e-mock runs on
manual dispatch and on release tags, where release-preflight requires
it. The paid lanes keep their `!= 'schedule'` guards in case a schedule
returns.

The tests that pinned the cron now pin its absence, and the build/CI
chapters describe the lane's real triggers. The same change goes to
`main` as a CI-only hotfix (owner decision 2C) so its stale schedule
stops starting the paid live stand before this release is promoted.
2026-10-05 17:21:52 +03:00
..
benchmarks Preserve OSWorld pre-admission audit reach after extraction 2026-10-03 18:12:51 +03:00
e2e_live CI: run no workflow on a schedule; the keyless E2E runs on dispatch and tags 2026-10-05 17:21:52 +03:00
__init__.py feat(devtools-benchmarks): add official benchmark harnesses and workspace executor 2026-06-06 12:03:30 +03:00
measure_review_pack.py fix: make review delivery and post-delivery evidence truthful 2026-09-27 21:53:44 +03:00
README.md Retire completed campaign artifacts and keep repository reports explicit 2026-09-18 01:07:51 +03:00

Ouroboros Devtools

devtools/ contains operator-side and benchmark support code that should be versioned with Ouroboros without becoming part of the runtime core.

Rules:

  • Generated logs, datasets, run outputs, Docker layers, and secrets do not live here.
  • Choose a benchmark output root outside the source checkout and runtime data; use each runner's documented output option or OUROBOROS_BENCH_RUNS_ROOT.
  • Runtime modules must not import devtools.
  • This is not an immune-system bypass: touched files are reviewed normally.
  • Promote code out of devtools only through a separate reviewed runtime plan.