The Agent Skills Directory

[COMMAND_EXECUTION]: The skill utilizes the bun runtime to execute its internal rendering and validation scripts, such as Tools/RenderClassic.ts and Tools/ValidateClassic.ts. This is a standard and safe implementation of the skill's core functionality.
[EXTERNAL_DOWNLOADS]: Fetches the playwright library and the Chromium browser engine from trusted official sources (NPM and Microsoft) to perform HTML-to-PNG conversion.
[PROMPT_INJECTION]: The skill processes classical Chinese text from user input or the web, creating a surface for indirect prompt injection.
Ingestion points: Reads text from user input and remote URLs in Workflows/AnnotateAndRender.md.
Boundary markers: None. Untrusted text is processed without explicit delimiters or instructions to ignore embedded commands.
Capability inventory: Performs file writes to the documents directory and executes local scripts via bun.
Sanitization: Employs escapeHtml in Tools/RenderClassic.ts to sanitize content before rendering it into the final image.
[SAFE]: The skill accesses configuration files of related skills by the same author (e.g., ljg-plain) located in the user's home directory (~/.agents/skills/). This cross-skill access is documented as part of the author's integrated toolset and does not pose a security risk.

ljg-classic