Conversation
Co-Authored-By: Claude Sonnet 4.6 <noreply@anthropic.com>
Test Results 335 files ±0 335 suites ±0 5m 4s ⏱️ +2s Results for commit 7bb6455. ± Comparison against base commit bdd53e6. This pull request removes 196 and adds 172 tests. Note that renamed tests count towards both.♻️ This comment has been updated with latest results. |
|
Keeping this in draft for now, I want to give this a test, even though the downside risk is minimal |
| @@ -1,3 +1,4 @@ | |||
| .claude/worktrees/ | |||
There was a problem hiding this comment.
And adding a slightly off topic change: I've found Claude & git worktrees very handy for multiple PRs at the same time
|
An idea: we don't read our performance pipeline output often, and until now I was manually reading a JSON diff whenever we had a "big" change worth measuring! What if we had an agentic pipeline constantly monitoring for performance regressions? We can also have a pipeline to explain errors, although everyone can click through the build output, it's handy to have the right paragraph already extracted. |
.github/workflows/ci-doctor.md
Outdated
| types: | ||
| - completed | ||
| branches: | ||
| - main |
Co-Authored-By: Claude Sonnet 4.6 <noreply@anthropic.com>
Co-Authored-By: Claude Sonnet 4.6 <noreply@anthropic.com>
Co-Authored-By: Claude Sonnet 4.6 <noreply@anthropic.com>
Co-Authored-By: Claude Sonnet 4.6 <noreply@anthropic.com>
Co-Authored-By: Claude Sonnet 4.6 <noreply@anthropic.com>
Adds the ci-doctor agentic workflow, which automatically investigates CI failures. This is one of GitHub's own agentic pipelines from their documentation.
When a monitored workflow fails, ci-doctor analyses failing builds and either comments on the PR or opens an issue (in the unlikely case of a master build failure)
It's not easy to test this as a dry-run. The agentic pipeline will only show up in the list of pipelines after merging into master. My plan is to give it a test on a deliberately failing pull request after this is merged into master.