The Agent Skills Directory

[INDIRECT_PROMPT_INJECTION]: The skill processes untrusted data such as HTML, SVG, and code snippets, rendering them using headless browsers (Playwright, Puppeteer) or CLI tools.
Ingestion points: The skill renders content provided via HTML strings, local files, or code files (e.g., source.py, content.html).
Boundary markers: No explicit delimiters or instructions are provided to the agent to ignore potentially malicious instructions embedded within the content to be rendered.
Capability inventory: The skill utilizes shell commands (npx, node, docker, silicon) and file-system writes to generate output images.
Sanitization: There is no evidence of sanitization, validation, or escaping of the external content before it is interpolated into rendering scripts or passed to CLI tools.
[COMMAND_EXECUTION]: The skill instructs the agent to execute various shell commands to perform rendering tasks.
Evidence includes the use of npx playwright, node -e for Puppeteer and Sharp, docker run for Mermaid, and CLI tools like silicon, inkscape, and convert.
[EXTERNAL_DOWNLOADS]: The skill references and utilizes external packages and services.
It suggests using npx to download and run @mermaid-js/mermaid-cli and playwright.
It provides a reference to the Carbon code-to-image service (carbon.now.sh).

exporting-to-png