skill-review

Pass

Audited by Gen Agent Trust Hub on Jul 30, 2026

Risk Level: SAFE
Full Analysis
  • [DATA_EXPOSURE]: The skill processes text from files within the .skill-proposals/ directory. It does not access sensitive system directories (e.g., .ssh, .aws), environment variables, or hardcoded credentials.
  • [COMMAND_EXECUTION]: The skill body contains no shell commands or scripts. The instructions explicitly state "edit nothing," ensuring that the agent remains in a read-only analytical state.
  • [EXTERNAL_DOWNLOADS]: There are no network requests, remote script fetches, or package installations defined in the skill or its evaluation fixtures.
  • [PROMPT_INJECTION]: Although the skill processes untrusted input (skill drafts), which is a common surface for indirect prompt injection, it lacks the tools or write permissions necessary to fulfill a malicious payload. Furthermore, it requires structured quoting of evidence, which serves as a natural delimiter.
Audit Metadata
Risk Level
SAFE
Analyzed
Jul 30, 2026, 04:48 PM
Security Audit — agent-trust-hub — skill-review