stuffed-toy-generator
Pass
Audited by Gen Agent Trust Hub on Aug 28, 2026
Risk Level: SAFECOMMAND_EXECUTIONINDIRECT_PROMPT_INJECTION
Full Analysis
- [COMMAND_EXECUTION]: The documentation provides examples using
curlandnpxto interact with the service. These commands execute network requests and run theonekeyagent tool, which is the intended interface for the skill. - [INDIRECT_PROMPT_INJECTION]: The skill processes text prompts and image URLs supplied by users through the
toy_generator_design_draftandtoy_generator_task_createfunctions. This content is sent to external 3D generation providers, creating a surface for potential injection attacks embedded within the source data. - Ingestion points:
promptandimagesparameters inSKILL.mdAPI specifications. - Boundary markers: Absent; there are no instructions to the agent to treat these specific fields as untrusted or to wrap them in delimiters.
- Capability inventory: The skill performs network operations via the OneKey Gateway (
agent.deepnlp.org) to submit tasks and retrieve results. - Sanitization: No explicit input validation or sanitization rules are documented for the text or URL inputs.
Audit Metadata