walkthrough
Pass
Audited by Gen Agent Trust Hub on Nov 12, 2026
Risk Level: SAFE
Full Analysis
- [SAFE]: The skill is designed for interactive code review and explicitly states that it is 'read-only except for an authorized walkthrough log.' It contains clear prohibitions against unauthorized commits, pushes, or deployments.
- [SAFE]: Instructions in
references/presenter.mdandreferences/walkthrough-log.mdenforce least-privilege principles, such as avoiding the storage of credentials, absolute machine paths, or the execution of destructive commands. - [SAFE]: The skill uses a
disable-model-invocation: trueflag in its frontmatter, which is a platform-level security control to restrict the skill's execution environment. - [SAFE]: While the skill suggests local exercises (e.g., running tests or checking UI), it mandates that these require existing authorization and must not have external effects without user consent.
Audit Metadata