model-integration

Pass

Audited by Gen Agent Trust Hub on Aug 28, 2026

Risk Level: SAFEINDIRECT_PROMPT_INJECTION
Full Analysis
  • [Indirect Prompt Injection Surface]: The skill is designed to ingest and process external content from "Reference repos" and "Inference scripts" provided by the user. While this is necessary for the skill's purpose, it creates a surface where the agent could potentially encounter malicious instructions embedded within the code or documentation of the reference implementation.
  • Ingestion points: The agent is instructed to ask for a GitHub link or local path to a reference repository (SKILL.md).
  • Boundary markers: The instructions do not currently include explicit guidance to treat external code as data only or to ignore embedded natural language instructions.
  • Capability inventory: The agent performs file system operations (creating model files, scripts, and documentation) and executes conversion and testing logic.
  • Sanitization: The workflow relies on the user to provide a legitimate reference repository and does not explicitly suggest sanitization steps for the ingested code.
Audit Metadata
Risk Level
SAFE
Analyzed
Aug 28, 2026, 11:52 AM
Security Audit — agent-trust-hub — model-integration