fix-issue

Pass

Audited by Gen Agent Trust Hub on Feb 23, 2026

Risk Level: SAFECOMMAND_EXECUTIONPROMPT_INJECTION
Full Analysis
  • [COMMAND_EXECUTION]: The skill uses npm test and git commit to perform its primary tasks. These are standard development operations and do not pose a threat in this context.\n- [PROMPT_INJECTION]: The skill is susceptible to indirect prompt injection through its ingestion of external data.\n
  • Ingestion points: The skill reads issue_id and error logs which could contain malicious instructions.\n
  • Boundary markers: No markers or safety instructions are used to delimit untrusted input from the system prompt.\n
  • Capability inventory: The agent can modify the local file system and execute shell commands for testing.\n
  • Sanitization: Input data is processed without escaping or validation against malicious content.
Audit Metadata
Risk Level
SAFE
Analyzed
Feb 23, 2026, 08:44 AM