The Agent Skills Directory

[EXTERNAL_DOWNLOADS]: The skill retrieves rendering engines and style assets from well-known, established providers. This includes library code from JSDelivr (Mermaid.js, Chart.js, Prism.js, and anime.js) and typefaces from Google Fonts. These downloads are required for the skill's primary function of generating high-quality standalone HTML documents.\n- [COMMAND_EXECUTION]: The skill uses development-oriented shell commands such as git diff, git show, ls, grep, and wc to gather the necessary metadata and file content from the local codebase. It also invokes system commands like open or xdg-open to display the resulting visualizations in the user's browser.\n- [COMMAND_EXECUTION]: The 'publish' mode utilizes a local shell script (publish.sh) located in a companion skill's directory. This is used to deploy the generated visualization to an external endpoint as intended by the author's workflow.\n- [PROMPT_INJECTION]: An indirect prompt injection surface exists because the skill processes untrusted codebase content and possesses significant system capabilities.\n
Ingestion points: The skill reads file content and git history (diff-review.md, fact-check.md) from the user's project into the agent context.\n
Boundary markers: Absent; the instructions do not explicitly provide delimiters or directives to the agent to disregard instructions that may be embedded within the source files it analyzes.\n
Capability inventory: The skill possesses command execution, file modification (in the fact-check prompt), and local network connectivity via a WebSocket bridge.\n
Sanitization: The instructions do not define any specific sanitization or filtering logic for the ingested content.

visual-explainer