uw-grill-layer
Pass
Audited by Gen Agent Trust Hub on Oct 30, 2026
Risk Level: SAFE
Full Analysis
- [SAFE]: The skill is designed for static analysis of codebases. It implements a strict 'read-only' policy for the source files and documentation it interrogates, ensuring that findings are evidence-backed through mandatory code quotes. The instructions explicitly prohibit the agent from interviewing users or modifying the primary documentation, which prevents race conditions and unauthorized changes in parallel execution environments.
- [INDIRECT_PROMPT_INJECTION]: The skill processes untrusted input in the form of source code and project documentation. This ingestion surface is inherent to its function as a code auditor. The risk of indirect prompt injection is mitigated by the skill's lack of network access (preventing data exfiltration) and the restricted set of shell tools (
ls,sed,grep) which are used exclusively for discovery and text processing without executing the analyzed content. - [COMMAND_EXECUTION]: The skill's configuration limits shell command execution to a safe subset of utilities. These tools are used for project navigation and searching, and the instructional logic reinforces that these should not be used to mutate the codebase.
Audit Metadata