The Agent Skills Directory

[PROMPT_INJECTION]: The skill is vulnerable to indirect prompt injection through a cross-site scripting (XSS) surface in the HTML generation logic. The renderHtml function in src/core/render.ts interpolates user-controlled chart options into a script block using JSON.stringify without escaping the closing script tag context.
Ingestion points: Data from the ChartRequest JSON object, including fields like title, subtitle, and dataset values.
Boundary markers: Absent; data is directly embedded into the HTML template without delimiters or safety instructions.
Capability inventory: The skill can read from and write to the local filesystem, including support for home and Desktop directories through CLI arguments handled in src/cli/args.ts.
Sanitization: Absent; the implementation relies solely on standard JSON stringification, which is insufficient for embedding data in HTML script tags.
[EXTERNAL_DOWNLOADS]: The generated HTML output references the ECharts library from a trusted and well-known service (https://cdn.jsdelivr.net/npm/echarts@5/dist/echarts.min.js). This is a neutral finding documented for visibility.
[COMMAND_EXECUTION]: The skill provides various CLI entry points (dist/cli/*.js) intended for agent execution. These tools allow the agent to manage charting workflows, which includes reading input data and saving generated artifacts to the filesystem.

echarts-chart-skill