writing-plans-conventions
Pass
Audited by Gen Agent Trust Hub on Aug 31, 2026
Risk Level: SAFEINDIRECT_PROMPT_INJECTION
Full Analysis
- [INDIRECT_PROMPT_INJECTION]: The skill processes external, untrusted 'spec or requirements' to generate coding plans, creating a surface for indirect prompt injection. * Ingestion points: The skill is triggered by user-provided task specifications as described in the SKILL.md frontmatter. * Boundary markers: No explicit delimiters or safety instructions are defined to separate untrusted specifications from the skill's own instructions. * Capability inventory: The generated plans involve shell command execution (build, type-check, lint, and test commands) and file system writes (creation of plan files in the project root). * Sanitization: The workflow relies on manual human-in-the-loop review (ExitPlanMode) rather than automated filtering or sanitization of the input requirements.
Audit Metadata