evals-clarify
Pass
Audited by Gen Agent Trust Hub on Nov 1, 2026
Risk Level: SAFECOMMAND_EXECUTIONINDIRECT_PROMPT_INJECTION
Full Analysis
- [COMMAND_EXECUTION]: Executes provided PowerShell and Bash scripts to resolve the project root and configuration. These scripts use standard system commands and the git CLI for environment discovery.
- [INDIRECT_PROMPT_INJECTION]: The skill ingests and transforms untrusted draft evaluation criteria, creating a surface where embedded instructions could influence agent behavior.
- Ingestion points: Processes lightweight drafts in Step 0 of SKILL.md.
- Boundary markers: Lacks explicit delimiters or ignore-instruction warnings for embedded content.
- Capability inventory: Writes to documentation files and JSON datasets, and triggers subsequent tool execution.
- Sanitization: No sanitization of draft content before template transformation.
Audit Metadata