manual-planning
Pass
Audited by Gen Agent Trust Hub on May 5, 2026
Risk Level: SAFE
Full Analysis
- [SAFE]: No malicious code, unauthorized network operations, or credential harvesting patterns were detected in the skill instructions or templates.
- [PROMPT_INJECTION]: The skill possesses an indirect prompt injection surface as it ingests repository context to populate plans (ingestion point: SKILL.md). While boundary markers and sanitization for this data are not explicitly defined, this processing is the primary function of the skill and does not constitute a direct threat.
- [COMMAND_EXECUTION]: The workflow involves generating validation commands for tasks, but these are constructed by the agent during runtime based on specific coding tasks and are not hardcoded malicious payloads.
Audit Metadata