deploying-machine-learning-models

Pass

Audited by Gen Agent Trust Hub on May 5, 2026

Risk Level: SAFEPROMPT_INJECTIONEXTERNAL_DOWNLOADS
Full Analysis
  • [PROMPT_INJECTION]: The skill facilitates an indirect prompt injection surface by analyzing user-provided model files and deployment configurations to automate its workflow. Malicious instructions embedded in the metadata or structure of these external assets could potentially influence the agent's behavior. Ingestion points: User-provided model files and deployment configurations described in SKILL.md. Boundary markers: No specific markers or delimiters are defined in the instructions to isolate external data from the execution context. Capability inventory: The skill is configured to use Bash, Write, Edit, Read, Grep, and Glob tools for infrastructure and file management (SKILL.md). Sanitization: No procedures for sanitizing or validating the content of external models are described in the provided documentation.
  • [EXTERNAL_DOWNLOADS]: The deploy_model.py script is documented to perform environment setup and dependency installation, which involves fetching packages from external registries. This is a standard pattern for deployment tools but represents a potential supply-chain surface area if dependencies are not strictly verified.
Audit Metadata
Risk Level
SAFE
Analyzed
May 5, 2026, 02:37 PM
Security Audit — agent-trust-hub — deploying-machine-learning-models