The Agent Skills Directory

[INDIRECT_PROMPT_INJECTION]: The skill processes untrusted user data which could contain malicious instructions designed to influence the AI agent's behavior during the 3D generation workflow.
Ingestion points: The prompt and images parameters in toy_generator_design_draft and toy_generator_task_create APIs accept user-supplied content (SKILL.md).
Boundary markers: The skill documentation does not specify the use of delimiters or instructions to ignore embedded commands within the prompt or image metadata.
Capability inventory: The skill performs network operations via curl and npx to external endpoints at agent.deepnlp.org (SKILL.md).
Sanitization: There is no evidence of sanitization or validation of the input strings before they are incorporated into API requests.
[EXTERNAL_DOWNLOADS]: The skill interacts with external services and fetches assets from remote domains.
The skill sends data to and receives results from https://agent.deepnlp.org/agent_router.
The skill references and displays images from static.aiagenta2z.com and us-static.aiagenta2z.com, which are used to store and preview generated 3D assets.

architecture-generator