The Agent Skills Directory

[EXTERNAL_DOWNLOADS]: The skill interacts with the generativelanguage.googleapis.com API to generate images. This is an interaction with a well-known service provided by Google for AI model inference.
[COMMAND_EXECUTION]: The skill provides a shell script in Phase 4 that utilizes curl to send requests to the Gemini API and pipes the output to a python3 one-liner. The Python snippet parses the JSON response, decodes the base64 image data, and writes the resulting PNG file to a specific local path (apps/web/public/images/books/). This is standard automation for handling API responses.
[INDIRECT_PROMPT_INJECTION]: The skill is designed to ingest and analyze untrusted data, specifically book chapters and outlines, to generate design thinking and image prompts.
Ingestion points: In Phase 1, the skill reads external book.yaml files, book outlines, and chapter text.
Boundary markers: None are specified to delimit user content or instruct the model to ignore embedded instructions.
Capability inventory: The skill has the capability to perform network requests via curl and write files to the local filesystem via python3.
Sanitization: There is no explicit sanitization or filtering of the input text before it is used to generate the cover design analysis and final prompt.

arcanea-book-cover