tool-use-guardian

Pass

Audited by Gen Agent Trust Hub on Jul 14, 2026

Risk Level: SAFEPROMPT_INJECTIONEXTERNAL_DOWNLOADS
Full Analysis
  • [EXTERNAL_DOWNLOADS]: The skill references resources and documentation from the Genesis Marketplace and API hosted on Vercel (genesis-node-api.vercel.app, genesis-marketplace.vercel.app), which is a well-known hosting service.
  • [PROMPT_INJECTION]: The skill is designed to intercept and process outputs from other tools to fix malformed data or handle errors. This creates a surface for Indirect Prompt Injection.
  • Ingestion points: Tool outputs (truncated JSON, error messages, and schema-mismatched data) from wrapped tool calls in SKILL.md.
  • Boundary markers: None identified in the provided documentation to distinguish between raw tool output and system instructions.
  • Capability inventory: Performs retries, auto-coercion of schemas, and resumes broken multi-step chains.
  • Sanitization: Documentation mentions 'auto-coercion' for schema mismatches, but does not specify validation or filtering of content before re-insertion into the agent context.
Audit Metadata
Risk Level
SAFE
Analyzed
Jul 14, 2026, 10:58 AM
Security Audit — agent-trust-hub — tool-use-guardian