land-branch
Pass
Audited by Gen Agent Trust Hub on Oct 23, 2026
Risk Level: SAFE
Full Analysis
- [SAFE]: The skill implements a comprehensive safety framework in
passive-data-safety.md, instructing the agent to disregard instruction-like content found in repository data such as diffs and commit messages, effectively mitigating indirect prompt injection risks. - [SAFE]: Automated secret detection and redaction (e.g.,
[redacted:api-key]) are mandatory for any content moved into pull request descriptions or commit bodies. - [SAFE]: Command execution is restricted to local git and GitHub CLI operations. The skill explicitly avoids shell interpolation when handling PR text, protecting against command injection.
- [SAFE]: Local scripts for readiness checks (
ship-guard.shandowner-notice.py) perform static analysis on the local file tree and metadata without any network activity or remote code execution. - [SAFE]: The skill enforces user confirmation for destructive actions, requiring literal matching of confirmation words like 'discard'.
Audit Metadata