task-brief-creator
Pass
Audited by Gen Agent Trust Hub on Oct 9, 2026
Risk Level: SAFEINDIRECT_PROMPT_INJECTION
Full Analysis
- [SAFE]: No malicious patterns or security vulnerabilities were detected. The skill operates by analyzing codebase context and user requirements to generate Markdown work plans in a designated directory.
- [INDIRECT_PROMPT_INJECTION]: The skill contains an inherent surface for indirect prompt injection as it ingests untrusted user input (e.g., PM notes, PRDs) and incorporates it into instructions for downstream agents. However, the skill implements a comprehensive evidence chain and mitigation strategy: 1. Ingestion points: User chat input and pasted planning documents (Stage 1). 2. Boundary markers: Strict enforcement of '[confirmed]'/'[inferred]' labels for evidence and an 'executable, not discursive' prose style rule. 3. Capability inventory: Limited to codebase analysis, file writing within the project, and execution of local validation scripts. 4. Sanitization: Multi-layered verification including structural validation via Python scripts, a content-level self-check (Stage 5.6), and downstream sub-agent verification loops (Stages 5.5 and 5.7).
Audit Metadata