skill-official-evaluation

Pass

Audited by Gen Agent Trust Hub on Jun 23, 2026

Risk Level: SAFE
Full Analysis
  • [COMMAND_EXECUTION]: The skill executes a local Python script (scripts/official_evaluate.py) to automate technical checks such as frontmatter parsing, directory structure validation, and regex-based secret scanning.
  • [PROMPT_INJECTION]: The skill possesses a surface for indirect prompt injection as it is designed to ingest and process untrusted content from the skill files it evaluates.
  • Ingestion points: File contents from the target skill (SKILL.md, scripts, and references) are read by the official_evaluate.py script and processed by the agent.
  • Boundary markers: Absent; the agent evaluates raw text artifacts without explicit isolation delimiters or 'ignore' instructions for embedded data.
  • Capability inventory: The skill is capable of directory traversal, file system read access, and generating evaluation reports (output to stdout or a local file).
  • Sanitization: No content sanitization is performed on the analyzed text; the agent relies on its internal instructions to apply the official rubric and verify artifacts as a mitigation against misleading content.
Audit Metadata
Risk Level
SAFE
Analyzed
Jun 23, 2026, 04:05 PM
Security Audit — agent-trust-hub — skill-official-evaluation