uw-help
Pass
Audited by Gen Agent Trust Hub on Oct 30, 2026
Risk Level: SAFECOMMAND_EXECUTIONEXTERNAL_DOWNLOADSINDIRECT_PROMPT_INJECTION
Full Analysis
- [COMMAND_EXECUTION]: The skill documentation specifies that it automatically executes build commands, specifically 'pnpm install && pnpm build', via the 'ensure_unwind_core' function during initial setup.
- [EXTERNAL_DOWNLOADS]: The skill's automated build process requires downloading and installing external Node.js packages, notably '@unwind/core'.
- [INDIRECT_PROMPT_INJECTION]: The skill processes external, untrusted source code as its primary input, creating a surface for indirect prompt injection attacks where malicious comments or code could influence agent behavior.
- Ingestion points: Source code files across multiple languages (TypeScript, JavaScript, Python, Rust, Java, C#) are ingested into the agent's context using Read, Grep, and Glob tools.
- Boundary markers: There are no documented boundary markers or system instructions provided to help the agent distinguish between code structure and potentially malicious embedded natural language instructions.
- Capability inventory: The workflow includes powerful capabilities such as 'uw-build' for reconstructing codebases in a target stack and various filesystem search tools.
- Sanitization: The instructions do not describe any sanitization, filtering, or validation steps for the content extracted from the analyzed codebase.
Audit Metadata