skills/frizzle-chan/mudd/add-check/Gen Agent Trust Hub

add-check

Pass

Audited by Gen Agent Trust Hub on Feb 17, 2026

Risk Level: SAFE
Full Analysis
  • [Indirect Prompt Injection] (LOW): The skill defines templates for modifying CI/CD and build tool configurations (justfile, GitHub Actions, Lefthook). While the skill itself is passive documentation, an agent using these templates with unvalidated external input for the YOUR_COMMAND placeholder could inadvertently introduce malicious commands into the build pipeline. Mandatory Evidence Chain:\n
  • Ingestion points: Instructions for the agent to take YOUR_COMMAND from user/external context and insert into config files.\n
  • Boundary markers: None present in the templates.\n
  • Capability inventory: File-write access to sensitive configurations (.github/workflows/ci.yaml, justfile, lefthook.yml).\n
  • Sanitization: None specified; relies on the agent's internal safety filters.\n- [Malicious Patterns] (SAFE): No evidence of prompt injection, obfuscation, or data exfiltration attempts.\n- [Privilege Escalation] (SAFE): While the skill involves modifying high-privilege files (CI workflows), it does so within the context of a developer tool and does not attempt to bypass system permissions or use sudo.
Audit Metadata
Risk Level
SAFE
Analyzed
Feb 17, 2026, 06:33 PM