skill-dependency-mapper

Pass

Audited by Gen Agent Trust Hub on Feb 18, 2026

Risk Level: SAFEPROMPT_INJECTIONCOMMAND_EXECUTION
Full Analysis
  • [PROMPT_INJECTION] (LOW): The skill processes metadata (names, descriptions, tags) from other skills in the environment. This ingestion point is a surface for indirect prompt injection where malicious metadata in another skill could influence agent behavior. Evidence: (1) Ingestion points: analyzer.scan_skills() reads files in /mnt/skills. (2) Boundary markers: Absent. (3) Capability inventory: Python execution of local scripts and filesystem writes to /tmp. (4) Sanitization: Absent.
  • [COMMAND_EXECUTION] (SAFE): The skill executes its own internal Python scripts (analyze_skills.py, detect_bottlenecks.py) as part of its core workflow. This is standard behavior for a diagnostic tool and does not involve remote code execution or suspicious commands.
Audit Metadata
Risk Level
SAFE
Analyzed
Feb 18, 2026, 12:02 AM