carpenter
Pass
Audited by Gen Agent Trust Hub on Oct 16, 2026
Risk Level: SAFEINDIRECT_PROMPT_INJECTION
Full Analysis
- [INDIRECT_PROMPT_INJECTION]: The skill processes external 'Architect Blueprints' and 'source material' to generate its drafts. This ingestion of untrusted or external data represents a theoretical surface for indirect prompt injection where instructions embedded in the source material could attempt to influence the agent's behavior. However, the skill's highly constrained role and structured output format mitigate this risk.
- Ingestion points: Architect Blueprint and source material/evidence (referenced in SKILL.md).
- Boundary markers: Absent; the skill does not explicitly define delimiters for external content.
- Capability inventory: Local file writing (draft files) and user interaction (AskUserQuestion).
- Sanitization: Not specified; the skill assumes the blueprint is already approved by the user.
- [SAFE]: The skill does not perform any sensitive network operations, access restricted system files, or execute arbitrary commands. All identified behaviors align with its stated purpose of prose construction and document management.
Audit Metadata