thinking-model-router

Pass

Audited by Gen Agent Trust Hub on Aug 17, 2026

Risk Level: SAFE
Full Analysis
  • [SAFE]: The skill consists entirely of instructional markdown for task categorization and routing logic. No executable code, scripts, or binaries are included.
  • [SAFE]: No network operations, file system modifications, or credential harvesting patterns were detected.
  • [INDIRECT_PROMPT_INJECTION]: The skill processes user-provided task descriptions to determine routing.
  • Ingestion points: User input describing a task or problem is evaluated against the provided procedure in SKILL.md.
  • Boundary markers: The skill includes instructions to short-circuit and exit if no clear fit is found, reducing the likelihood of forced or malicious routing.
  • Capability inventory: The skill only outputs text strings (slugs) intended for use with a Skill tool; it lacks any system-level capabilities like subprocess execution or network access.
  • Sanitization: The procedure explicitly instructs the agent to use only cited IDs and avoid inventing slugs, providing a narrow allowlist for output.
Audit Metadata
Risk Level
SAFE
Analyzed
Aug 17, 2026, 03:56 PM
Security Audit — agent-trust-hub — thinking-model-router