test: line-anchored audit demo#8
Conversation
Same semantic — still vague about output artifact — so CONTENT-005 will still fire. Now line 3 is in the PR diff, exercising the line-anchored review-comment path. Co-Authored-By: Claude Opus 4.7 (1M context) <noreply@anthropic.com>
|
2027 auto-runs evals against preview deployments of your docs. To enable this, install one of:
Once a preview is deployed, open a new PR and we'll run the eval automatically. Evaluating agent experience using 2027.dev · View dashboard |
![2027-evals[bot]](https://avatars.githubusercontent.com/in/3339941?s=60&v=4){kind=small}
| --- | ||
| name: autobrowse | ||
| description: Self-improving browser automation via the auto-research loop. Iteratively runs a browsing task, reads the trace, and improves the navigation skill (strategy.md) until it reliably passes. Supports parallel runs across multiple tasks using sub-agents. Use when you want to build or improve browser automation skills for specific website tasks. | ||
| description: Iterative browser automation that improves itself by reading traces. Runs a browsing task, analyzes what went wrong, and tunes the navigation strategy until the task passes reliably. Supports parallel runs across multiple tasks. Use when you want to build or refine browser automation for specific website tasks. |
There was a problem hiding this comment.
🟡 Description never states what concrete artifact or return value is produced; it only explains the iterative process and purpose without specifying output format, file path, committed change, or deployed service.
There was a problem hiding this comment.
🟡 check threw: Failed after 3 attempts. Last error: Your account org-e0ffd1b77a2e4291830bf49c6a4bff07 request reached organization max RPM: 20, please try again after 1 seconds
There was a problem hiding this comment.
🟡 check threw: Failed after 3 attempts. Last error: Organization Rate limit exceeded, please try again after 1 seconds
2 participants
Touches line 3 of autobrowse/SKILL.md (the description). Used to verify the audit posts a real line-anchored review comment (not file-level fallback) when the finding's line is in the PR diff.