catchup

Pass

Audited by Gen Agent Trust Hub on Feb 17, 2026

Risk Level: SAFEPROMPT_INJECTIONCOMMAND_EXECUTION
Full Analysis
  • Indirect Prompt Injection (LOW): The skill has a potential surface for indirect prompt injection by processing external codebase data that could contain malicious instructions.
  • Ingestion points: Output from git diff master and files in specs/.
  • Boundary markers: Not present in the instructions.
  • Capability inventory: The skill has access to Bash(git *), Read, and Glob tools.
  • Sanitization: No sanitization is performed on the ingested content.\n- Command Execution (SAFE): The skill utilizes the Bash(git *) tool to execute git commands. This is consistent with the skill's primary purpose of reviewing codebase history and changes, and the tool is restricted to the git binary.
Audit Metadata
Risk Level
SAFE
Analyzed
Feb 17, 2026, 06:24 PM