An agent skill that turns complex terminal output into styled HTML pages you actually want to read.
Ask your agent to explain a system architecture, review a diff, or compare requirements against a plan. Instead of ASCII art and box-drawing tables, it generates a self-contained HTML page and opens it in your browser:
> draw a diagram of our authentication flow
> /diff-review
> /plan-review ~/docs/refactor-plan.md
Each one produces a single .html file with real typography, dark/light theme support, and interactive Mermaid diagrams with zoom and pan. No build step, no dependencies beyond a browser.
visual-explainer.mp4
Every coding agent defaults to ASCII art when you ask for a diagram. Box-drawing characters, monospace alignment hacks, text arrows. It works for trivial cases, but anything beyond a 3-box flowchart turns into an unreadable mess that nobody would put in a presentation or share with a team.
Tables are worse. Ask the agent to compare 15 requirements against a plan and you get a wall of pipes and dashes that wraps and breaks in the terminal. The data is there but it's painful to read.
The skill follows the Agent Skills specification. Clone it into your agent's skills directory:
# Pi
git clone https://github.com/nicobailon/visual-explainer.git ~/.pi/agent/skills/visual-explainer
# Claude Code
git clone https://github.com/nicobailon/visual-explainer.git ~/.claude/skills/visual-explainer
# Other agents — point at the directory containing SKILL.md,
# or paste its contents into your system promptFor Pi, restart after cloning. To get the slash commands (/diff-review, /plan-review, etc.), copy the prompt templates:
cp ~/.pi/agent/skills/visual-explainer/prompts/*.md ~/.pi/agent/prompts/If you have surf-cli installed, the skill can also generate illustrations via Gemini Nano Banana Pro and embed them in pages. The agent detects surf automatically and skips image generation if it's not there.
The agent loads the skill when you mention diagrams, architecture, flowcharts, schemas, or visualizations. It also kicks in automatically when it's about to dump a complex table in the terminal (4+ rows or 3+ columns) — it renders HTML instead and opens it in the browser. Output goes to ~/.agent/diagrams/.
The skill ships with five prompt templates:
| Command | What it does |
|---|---|
/generate-web-diagram |
Generate an HTML diagram for any topic |
/diff-review |
Visual diff review with architecture comparison, code review, decision log |
/plan-review |
Compare a plan against the codebase with risk assessment |
/project-recap |
Mental model snapshot for context-switching back to a project |
/fact-check |
Verify accuracy of a review page or plan doc against actual code |
/diff-review is probably the most useful. Run it with no arguments to diff against main, or pass any git ref:
/diff-review # feature branch vs main (default)
/diff-review abc123 # single commit
/diff-review main..HEAD # committed changes only
/diff-review #42 # pull request
It generates a full page with before/after architecture diagrams, KPI dashboard, structured Good/Bad/Ugly code review, decision log with confidence indicators, and re-entry context for your future self.
/plan-review does something similar but for implementation plans — pass it a plan file and it cross-references every claim against the actual codebase, produces current vs. planned architecture diagrams, and flags risks and gaps:
/plan-review ~/docs/refactor-plan.md
/project-recap is designed for context-switching back to a project after days away. It scans recent git activity and produces an architecture snapshot, decision log, and cognitive debt hotspots. /fact-check takes any document that makes claims about code and verifies every one of them.
SKILL.md (workflow + design principles)
↓
references/ ← agent reads before each generation
├── css-patterns.md (layouts, animations, theming, depth tiers)
├── libraries.md (Mermaid theming, Chart.js, anime.js, font pairings)
└── responsive-nav.md (sticky sidebar TOC for multi-section pages)
↓
templates/ ← agent reads the matching reference template
├── architecture.html (CSS Grid cards — terracotta/sage palette)
├── mermaid-flowchart.html (Mermaid + ELK + handDrawn — teal/cyan palette)
└── data-table.html (tables with KPIs and badges — rose/cranberry palette)
↓
~/.agent/diagrams/filename.html → opens in browser
The agent picks an aesthetic direction, reads the right reference template, generates a self-contained HTML file with both light and dark themes, and opens it. The three templates use deliberately different palettes so the agent learns variety rather than defaulting to one look. The skill handles 11 diagram types — Mermaid for anything with connections (flowcharts, sequences, ER, state machines, mind maps), CSS Grid for text-heavy architecture overviews, HTML tables for data, Chart.js for dashboards — and routes to the right approach automatically.
To customize the output directory, browser command, or add your own diagram types and CSS patterns, edit the files directly. The agent reads them fresh each time.
- Requires a browser to view — no inline terminal rendering
- Switching OS theme requires a page refresh for Mermaid SVGs (CSS-styled elements respond instantly)
- Results vary by model capability — the skill provides design guidance, not pixel-perfect specs
Borrows ideas from Anthropic's frontend-design skill and interface-design, adapted for one-shot diagram generation.
MIT