woostack-ask
Pass
Audited by Gen Agent Trust Hub on Jul 31, 2026
Risk Level: SAFEPROMPT_INJECTION
Full Analysis
- [SAFE]: The skill enforces a comprehensive 'WRITE-BLOCK' policy, ensuring that no modifications are made to the codebase, configuration files, or version control system during the investigation process.
- [PROMPT_INJECTION]: The skill defines a surface for indirect prompt injection through its ingestion of external data and project management artifacts. 1. Ingestion points: External content is pulled via web search/fetch tools, and remote artifact text is retrieved from Linear via internal adapter scripts (SKILL.md). 2. Boundary markers: Strong instructional boundaries are present; the skill explicitly mandates that all remote data must be treated as untrusted evidence and forbids it from directing tools or altering the command's execution flow. 3. Capability inventory: Capabilities are restricted to read-only repository access (Grep, Glob, Read) and external data retrieval. 4. Sanitization: The skill relies on explicit agent instructions to distinguish between evidence and instructions rather than automated content sanitization.
- [SAFE]: External network operations are strictly limited to data retrieval; instructions explicitly forbid the exfiltration of repository content to external domains.
- [SAFE]: The skill utilizes local project scripts for backend resolution and artifact management, maintaining a predictable execution environment within the project's established structure.
Audit Metadata