skill-init

Pass

Audited by Gen Agent Trust Hub on Jun 27, 2026

Risk Level: SAFE
Full Analysis
  • [PROMPT_INJECTION]: The skill presents an indirect prompt injection surface by ingesting user-provided professional details and storing them in a persistent state file for later consumption by other skills. \n
  • Ingestion points: Phase 2 user questionnaire responses including current skills and goals. \n
  • Boundary markers: No specific delimiters or "ignore instructions" warnings are applied to the user-supplied data in the state file. \n
  • Capability inventory: The skill utilizes Read, Write, Bash, Edit, Glob, and Skill tools. \n
  • Sanitization: No input sanitization, escaping, or validation is implemented before writing data to the .skill-state.json file. \n- [COMMAND_EXECUTION]: Benign use of shell commands to verify the existence of the state file and to query system time and timezone settings. \n
  • Evidence: The skill uses test -f, date, and readlink /etc/localtime commands to initialize its environment and record metadata.
Audit Metadata
Risk Level
SAFE
Analyzed
Jun 27, 2026, 06:20 AM
Security Audit — agent-trust-hub — skill-init