fix-loop

Pass

Audited by Gen Agent Trust Hub on Apr 12, 2026

Risk Level: SAFECOMMAND_EXECUTIONPROMPT_INJECTION
Full Analysis
  • [COMMAND_EXECUTION]: The protocol employs standard version control and search utilities, specifically git and grep, to stage changes and verify fixes. These operations are logically restricted to the files associated with confirmed review findings.- [PROMPT_INJECTION]: The skill ingests structured 'Finding' data which contains natural language fields such as body and evidence. This creates an indirect prompt injection surface where instructions embedded in a malicious review could attempt to influence the subagents performing the fixes.
  • Ingestion points: Findings processed by the protocol as defined in SKILL.md.
  • Boundary markers: Absent; the skill does not specify delimiters or instructions for the subagents to ignore content within the finding fields.
  • Capability inventory: Local file modification (subagents) and test execution capabilities mentioned in SKILL.md.
  • Sanitization: Absent; the findings are passed to subagents based on their conformance to a schema, but without textual sanitization of description fields.
Audit Metadata
Risk Level
SAFE
Analyzed
Apr 12, 2026, 02:15 PM
Security Audit — agent-trust-hub — fix-loop