Visual-explainer is an agent skill that transforms terminal output into styled HTML pages and slide decks for diagrams, diffs, audits, tables, and project summaries.
The tool addresses the problem that coding agents default to ASCII art and box-drawing characters when asked to visualize information, which becomes unreadable beyond simple cases. Tables comparing multiple items produce walls of pipes and dashes that wrap and break in the terminal. Visual-explainer intercepts these requests and generates self-contained HTML pages with real typography, dark and light theme support, and interactive Mermaid diagrams featuring zoom and pan capabilities. The skill requires no build step and no dependencies beyond a browser for normal use.
The tool suits developers working with AI agents who want readable visual output for architecture diagrams, code diffs, requirement comparisons, and data tables. It integrates with multiple agent platforms including Claude Code via marketplace plugin, Pi, MCP hosts as a local stdio server, and native skill paths for Antigravity CLI, Codex CLI, OpenCode, Cursor, and OpenClaw. A separate utility converts simple HTML slide decks to PPTX format, though HTML remains the source of truth. Installation varies by platform but generally involves copying the skill directory to the appropriate location or installing through package metadata.
The project shows active development with support across multiple agent harnesses and platforms, indicating ongoing maintenance to keep pace with different tool ecosystems. The codebase includes optional Node dependencies for MCP and PPTX utilities while maintaining a zero-dependency core for browser-based rendering. Documentation covers installation paths for each supported platform with specific guidance on namespace conventions and configuration locations.
