self-review

Pass

Audited by Gen Agent Trust Hub on Aug 28, 2026

Risk Level: SAFEINDIRECT_PROMPT_INJECTIONCOMMAND_EXECUTION
Full Analysis
  • [Command Execution]: The skill utilizes standard git commands such as git diff, git log, and git show to retrieve code changes for analysis. These operations are routine for development workflows and are used here to inform the review process.\n- [Indirect Prompt Injection]: The skill processes external code changes, which presents a potential surface for indirect instructions embedded within the code or comments.\n
  • Ingestion points: Code changes are ingested via git diff and git show commands as specified in SKILL.md.\n
  • Boundary markers: The instructions do not define specific delimiters to isolate the diff content from the agent's logic.\n
  • Capability inventory: The skill is limited to read operations and local git commands; it explicitly excludes file modification and network connectivity.\n
  • Sanitization: No specific content sanitization or escaping mechanisms are described for the diff output prior to analysis.
Audit Metadata
Risk Level
SAFE
Analyzed
Aug 28, 2026, 11:53 AM
Security Audit — agent-trust-hub — self-review