jinko-solution-and-product-guide

Pass

Audited by Gen Agent Trust Hub on Sep 4, 2026

Risk Level: SAFE
Full Analysis
  • [EXTERNAL_DOWNLOADS]: The skill fetches a JSON model library from https://doc.jinko.ai/model-library.json. This is a vendor-controlled domain (Nova In Silico / Jinkō) and the operation is documented as a primary function for retrieving catalog metadata.
  • [COMMAND_EXECUTION]: The skill executes a local Python script scripts/search_model_library.py. This script is part of the skill package and implements defensive measures including size limits on responses, strict JSON schema validation, and host pinning to prevent SSRF or unexpected redirects.
  • [INDIRECT_PROMPT_INJECTION]: The skill ingests data from the external model-library.json file. The risk is mitigated by strict field validation (ensuring all fields are strings) and the use of the data for informational catalog purposes only, without feeding it into high-privilege execution sinks.
  • [DYNAMIC_EXECUTION]: The skill uses json.loads to parse the fetched library. The implementation includes type checking and bounds checking to ensure the data adheres to the expected scientific model schema.
Audit Metadata
Risk Level
SAFE
Analyzed
Sep 4, 2026, 11:58 AM
Security Audit — agent-trust-hub — jinko-solution-and-product-guide