trace-it

Pass

Audited by Gen Agent Trust Hub on Feb 17, 2026

Risk Level: SAFE
Full Analysis
  • COMMAND_EXECUTION (SAFE): The skill utilizes bash commands grep, find, and git for local code analysis. These tools are industry standard for dependency tracing and do not involve arbitrary command execution or risky network requests.
  • DATA_EXPOSURE (SAFE): The skill reads local files in the src/ directory to identify function callers. There is no evidence of attempts to access sensitive system files (e.g., ~/.ssh) or transmit data externally.
  • INDIRECT_PROMPT_INJECTION (SAFE): The skill possesses a data ingestion surface by reading project source files, but the potential for malicious instructions is mitigated by a lack of destructive tools or network access.
  • Ingestion points: Project source files in src/ (referenced in SKILL.md).
  • Boundary markers: The skill provides structured markdown templates for the agent to document its analysis.
  • Capability inventory: grep, find, git, and file: read. No network, write, or privilege escalation capabilities are present.
  • Sanitization: While no explicit content sanitization is mentioned, the limited toolset prevents successful exploitation of injected instructions.
Audit Metadata
Risk Level
SAFE
Analyzed
Feb 17, 2026, 06:36 PM