The Agent Skills Directory

[COMMAND_EXECUTION]: The skill automates the creation and validation of MicroSims by executing local Python scripts from a defined utility directory (src/microsim-utils/).
Utilities include extract-sim-specs.py, generate-sim-scaffold.py, validate-sims.py, and update-mkdocs-nav.py.
The skill also utilizes a local system utility, bk-capture-screenshot, to generate preview images of the simulations using headless Chrome.
[EXTERNAL_DOWNLOADS]: Generated templates and guides include references to well-known JavaScript visualization libraries hosted on trusted Content Delivery Networks (CDNs).
Sources include jsdelivr.net, unpkg.com, and plot.ly.
Resources targeted are official distributions of p5.js, Chart.js, Plotly.js, Mermaid.js, vis-network, vis-timeline, and Leaflet.js.
[PROMPT_INJECTION]: The skill incorporates a vulnerability surface for indirect prompt injection by ingesting and processing untrusted data from chapter files.
Ingestion points: Chapter index.md files are parsed to extract simulation specifications (#### Diagram: headers).
Boundary markers: There are no specific markers or warnings instructed to the parsing logic to ignore embedded instructions within the user-provided markdown.
Capability inventory: The skill possesses file-write capabilities (creating .js and .html files) and the ability to execute local Python scripts and a screenshot tool.
Sanitization: The skill does not explicitly provide for sanitization of extracted summary or spec text before it is interpolated into documentation or metadata files.

microsim-generator