bisect-regression
Pass
Audited by Gen Agent Trust Hub on Jun 22, 2026
Risk Level: SAFE
Full Analysis
- [SAFE]: The skill provides structured instructions for performing a Git bisect to locate regressions, with explicit rules to protect user data and repository state.
- [SAFE]: Includes defensive checks such as the 'No-evidence rule' and 'Stop / Confirmation Rules' which prevent the agent from making unverified claims or performing destructive actions without confirmation.
- [SAFE]: No evidence of network exfiltration, hardcoded credentials, or suspicious file access was found in the skill or its references.
- [SAFE]: The skill does not employ any form of obfuscation or hidden logic.
- [SAFE]: While the skill executes local test commands, this is its primary intended function and is governed by strict requirements for isolated worktrees and deterministic signals.
Audit Metadata