batch-plan-validate
Pass
Audited by Gen Agent Trust Hub on Sep 1, 2026
Risk Level: SAFEINDIRECT_PROMPT_INJECTION
Full Analysis
- [INDIRECT_PROMPT_INJECTION]: The skill processes external data from roadmap files to automate the creation of PRDs and implementation plans, creating a surface for indirect instructions to influence the agent's behavior.
- Ingestion points: Top-level unchecked checklist items parsed from roadmap files and explicit task lists provided by the user in SKILL.md.
- Boundary markers: The skill lacks explicit instructions to treat roadmap content as untrusted or to use delimiters (e.g., XML tags or triple quotes) when interpolating this data into prompts for drafting skills.
- Capability inventory: The skill possesses significant capabilities including file system write access for creating PRDs and plans, invocation of research subagents (Explore/Plan), and the ability to trigger environment-level validation modes (EnterPlanMode/ExitPlanMode).
- Sanitization: No evidence of input validation, escaping, or filtering is present for the task labels or descriptions extracted from the roadmap source.
Audit Metadata