tune

Pass

Audited by Gen Agent Trust Hub on Aug 16, 2026

Risk Level: SAFEPROMPT_INJECTIONCOMMAND_EXECUTION
Full Analysis
  • [PROMPT_INJECTION]: The skill processes a friction log containing records of previous agent failures, which may include data derived from untrusted external sources, presenting an indirect prompt injection risk.\n
  • Ingestion points: The agent reads from .claude/issue-to-pr/friction.log (SKILL.md).\n
  • Boundary markers: A mandatory human approval step is required before applying any proposed changes (SKILL.md, Step 4).\n
  • Capability inventory: The skill can modify the SKILL.md file, scripts, and reference files, and executes a shell script tests/run-tests.sh (SKILL.md, Step 4).\n
  • Sanitization: No specific sanitization or filtering logic is described for the log content; it relies on the model's interpretation and human review.\n- [COMMAND_EXECUTION]: The skill executes repository-local shell scripts to verify proposed changes.\n
  • Evidence: The skill runs tests/run-tests.sh as part of its verification process (SKILL.md, Step 4).
Audit Metadata
Risk Level
SAFE
Analyzed
Aug 16, 2026, 09:36 PM
Security Audit — agent-trust-hub — tune