The Agent Skills Directory

[INDIRECT_PROMPT_INJECTION]: The skill is designed to ingest and process user-supplied message content and external image URLs, which represents a surface for indirect prompt injection attacks.
Ingestion points: The messages and image_urls parameters in the API request examples in SKILL.md accept external data.
Boundary markers: The documentation does not provide delimiters or specific instructions for the AI to ignore malicious commands that might be embedded in the user text or processed images.
Capability inventory: The skill utilizes the curl tool for network communication and requires access to the EACHLABS_API_KEY environment variable.
Sanitization: No validation, filtering, or sanitization logic is described for the user-provided inputs.\n- [COMMAND_EXECUTION]: The skill provides instructions for the agent to execute network commands to interact with external services.
Evidence: SKILL.md contains multiple examples of curl POST requests directed at https://eachsense-agent.core.eachlabs.run/v1/chat/completions for image generation, refinement, and product mockup creation.

pet-portrait-generation