skills/nasrulhazim/claude/self-update/Gen Agent Trust Hub

self-update

Pass

Audited by Gen Agent Trust Hub on Mar 3, 2026

Risk Level: SAFEPROMPT_INJECTION
Full Analysis
  • [PROMPT_INJECTION]: Indirect Prompt Injection Surface. The skill is designed to manage a persistent instruction file (CLAUDE.md) that is read by the agent to guide future actions. This creates an attack surface where malicious or unintended instructions can be persisted.
  • Ingestion points: The skill explicitly reads the CLAUDE.md file during the 'Update Procedure' in SKILL.md.
  • Boundary markers: No boundary markers or 'ignore' instructions are implemented to isolate the content of CLAUDE.md from the agent's system prompt.
  • Capability inventory: The skill possesses the capability to read and write to the local file system (specifically CLAUDE.md and its template).
  • Sanitization: There is no evidence of sanitization or filtering of user-provided preferences before they are written to the persistent document.
Audit Metadata
Risk Level
SAFE
Analyzed
Mar 3, 2026, 11:55 PM
Security Audit — agent-trust-hub — self-update