The Agent Skills Directory

[COMMAND_EXECUTION]: The skill instructs the agent to execute several local command-line tools for image conversion.
Tools include silicon, inkscape, convert (ImageMagick), termshot, and docker.
It uses docker run with volume mounting (-v "$(pwd)":/data) to run the mermaid-cli container, which provides access to the current working directory.
[EXTERNAL_DOWNLOADS]: The skill uses npx to fetch and execute packages at runtime.
It references @mermaid-js/mermaid-cli and playwright for rendering tasks. These are well-known and reputable packages used for their intended purpose.
[DYNAMIC_EXECUTION]: The skill utilizes node -e to execute inline JavaScript snippets.
These snippets use the puppeteer and sharp libraries to handle HTML rendering and SVG conversion. The logic is hardcoded and scoped to conversion tasks.
[INDIRECT_PROMPT_INJECTION]: The skill processes external data formats that could contain malicious instructions.
Ingestion points: Reads HTML, SVG, Code snippets, and Mermaid diagram definitions from the workspace (SKILL.md).
Boundary markers: None identified. Instructions do not specify delimiters for untrusted content.
Capability inventory: Executes shell commands via npx, node, and docker across multiple methods.
Sanitization: No evidence of input sanitization or validation before rendering or conversion. This surface could be exploited if an attacker provides malicious HTML/SVG designed to influence the agent if it processes the rendered output.

exporting-to-png