skills/amit-t/skills/prd-to-plan/Gen Agent Trust Hub

prd-to-plan

Pass

Audited by Gen Agent Trust Hub on Oct 31, 2026

Risk Level: SAFEINDIRECT_PROMPT_INJECTIONCOMMAND_EXECUTION
Full Analysis
  • [INDIRECT_PROMPT_INJECTION]: The skill is designed to process external PRD documents and use them to guide codebase exploration and file generation, which introduces a surface for malicious instructions to influence the agent's behavior.
  • Ingestion points: Ingests PRD data directly from the conversation or by reading user-pointed files as instructed in SKILL.md (Step 1).
  • Boundary markers: The instructions do not define clear delimiters or include warnings for the agent to ignore any embedded instructions within the PRD content.
  • Capability inventory: The skill directs the agent to 'explore the codebase' (which requires broad read access) and 'Write the plan as a Markdown file' to the ./plans/ directory (write access) as specified in SKILL.md (Steps 2 and 6).
  • Sanitization: There are no instructions for sanitizing the input data or validating generated file paths, which could be exploited if a PRD contains malicious payloads.
  • [COMMAND_EXECUTION]: To fulfill its purpose, the skill instructs the agent to 'explore the codebase' to understand architecture and integration layers. While necessary for the task, this involves execution of system commands (like file listing and reading) that could be co-opted if the agent is compromised via indirect injection.
Audit Metadata
Risk Level
SAFE
Analyzed
Oct 31, 2026, 04:28 PM