slice-work
Pass
Audited by Gen Agent Trust Hub on Sep 23, 2026
Risk Level: SAFE
Full Analysis
- [INDIRECT_PROMPT_INJECTION]: The skill is designed to ingest and process external requirements and current code to generate roadmaps. While this creates a surface for indirect prompt injection, the skill mitigates risk by explicitly limiting the agent's actions to documentation creation and forbidding high-privilege operations such as code execution, commits, or network deployment.
- Ingestion points: Reads user requests, external requirements, and existing repository code/tests (SKILL.md).
- Boundary markers: Instructs the agent to distinguish between authoritative sources and proposals, and to surface contradictions rather than resolving them automatically.
- Capability inventory: The skill is restricted to file writing (documentation only). It explicitly states: 'No commits, pushes, deployments, or new dependency installation are authorized'.
- Sanitization: None mentioned, but capabilities are sufficiently restricted.
Audit Metadata