The Agent Skills Directory

[INDIRECT_PROMPT_INJECTION]: The skill is designed to process and transform existing business cases, teaching notes, and research evidence which are essentially untrusted data sources.
Ingestion points: The skill reads case.md, teaching-note.md, linked exhibits, and research sources from the project workspace (specified in SKILL.md).
Boundary markers: The instructions do not define explicit delimiters or instructions to the agent to treat the ingested case content as data only, potentially allowing instructions embedded in those files to influence agent behavior.
Capability inventory: The skill can write files to the workspace and execute a local rendering script (scripts/render-reader-formats.ts) that spawns system processes.
Sanitization: The render-reader-formats.ts script implements basic HTML escaping (escapeHtml, escapeAttr) and a safeUrl filter when converting Markdown to HTML.
[COMMAND_EXECUTION]: The skill uses a local script (scripts/render-reader-formats.ts) to automate the generation of reading copies. The script uses the node:child_process module to spawn a headless Chrome/Chromium browser for PDF printing and diagram rendering, and it uses system openers (open, cmd /c start, or xdg-open) to display the results to the user. These operations are restricted to the local workspace and intended for the skill's primary functionality.
[EXTERNAL_DOWNLOADS]: The render-reader-formats.ts script fetches the Mermaid.js library from the jsDelivr CDN (https://cdn.jsdelivr.net/npm/mermaid@12.0.0/dist/mermaid.min.js) at runtime to render diagrams. This is a well-known service used for serving standard web libraries.

case-disguise