The Agent Skills Directory

[COMMAND_EXECUTION]: The skill instructs the agent to use shell utilities such as curl, base64, jq, file, and pngquant to manage the lifecycle of asset generation, from downloading to optimization.
[EXTERNAL_DOWNLOADS]: The skill fetches image files and JSON metadata from domains associated with the PixelLab service, specifically api.pixellab.ai and backblaze.pixellab.ai.
[DATA_EXFILTRATION]: The skill involves uploading local image data to an external API for processing tasks like animation and background removal.
Evidence: Shell commands in references/api-v2-extra.md demonstrate encoding local PNGs to Base64 and sending them to https://api.pixellab.ai/v2/ via POST requests.
[INDIRECT_PROMPT_INJECTION]: The skill processes structured and unstructured data from external API responses and interpolates user-provided descriptions into API requests, creating an ingestion surface for potentially untrusted content.
Ingestion points: JSON responses containing URLs and Base64-encoded image data from api.pixellab.ai are processed in files like SKILL.md and references/api-v2-extra.md.
Boundary markers: No explicit delimiters or safety warnings are provided for the interpolation of user-supplied descriptions into the action or description fields of API calls.
Capability inventory: The skill utilizes shell execution tools (curl, base64) and file system writes.
Sanitization: The instructions recommend validating file types using the file command after download to ensure the integrity of assets.

pixellab