write-prd
Pass
Audited by Gen Agent Trust Hub on Mar 13, 2026
Risk Level: SAFE
Full Analysis
- [SAFE]: The skill is dedicated to product documentation and strategy with no evidence of malicious logic or safety violations.
- [PROMPT_INJECTION]: No direct prompt injection or attempts to bypass agent safety guidelines were identified. The skill uses natural instructional language for template generation.
- [DATA_EXFILTRATION]: No hardcoded credentials, sensitive file access (e.g., SSH keys), or unauthorized data exfiltration patterns were found. File writing is scoped to PRD artifacts.
- [REMOTE_CODE_EXECUTION]: The skill does not perform remote code execution, script downloads, or unverified package installations.
- [COMMAND_EXECUTION]: While the Bash tool is authorized, its usage in the skill is limited to legitimate operations like writing files and updating task status. No arbitrary shell command patterns were detected.
- [PROMPT_INJECTION]: Indirect prompt injection surface detected via ingestion of product arguments and memory search results (SKILL.md); boundary markers and sanitization are absent, but the capabilities (Write, Bash) are used for intended document generation purposes.
Audit Metadata