guardian-creator

Pass

Audited by Gen Agent Trust Hub on Jul 25, 2026

Risk Level: SAFEPROMPT_INJECTION
Full Analysis
  • [PROMPT_INJECTION]: The skill is vulnerable to indirect prompt injection because it reads content from external markdown files and interpolates it into the output configuration file.
  • Ingestion points: The skill reads files from <repo-root>/ai-tools/command-briefs/ and <repo-root>/ai-tools/skills/ as defined in the 'Four-step workflow'.
  • Boundary markers: There are no protective boundary markers or instructions for the agent to ignore potentially malicious embedded commands within the source data.
  • Capability inventory: The skill utilizes file system read/write operations and external network tools like WebFetch and web_search_exa to gather context.
  • Sanitization: The instructions explicitly direct the agent to copy content such as 'Critical directives' verbatim from the source files into the generated subagent file, without any sanitization or validation of the input content.
Audit Metadata
Risk Level
SAFE
Analyzed
Jul 25, 2026, 03:22 AM
Security Audit — agent-trust-hub — guardian-creator