review-fix-codex
Pass
Audited by Gen Agent Trust Hub on Apr 11, 2026
Risk Level: SAFEEXTERNAL_DOWNLOADSCOMMAND_EXECUTIONPROMPT_INJECTION
Full Analysis
- [EXTERNAL_DOWNLOADS]: The skill downloads and executes the
@willbooster/agent-skillspackage from the NPM registry usingbunx. Since the package is owned by the skill author (WillBooster), this is a vendor-native dependency. - [COMMAND_EXECUTION]: The workflow involves executing shell commands through
bunxand performing Git operations like commit and push to apply code fixes. - [PROMPT_INJECTION]: The skill processes output from an external tool as instructions for modifying source code, which represents an indirect prompt injection surface. The risk is managed by the agent's instruction to verify comment validity and the use of trusted vendor tooling. Ingestion points: output from the review command in SKILL.md; Boundary markers: absent; Capability inventory: file system modification and Git push operations; Sanitization: absent.
Audit Metadata