By sidfeels
Generate visually excellent, hand-drawn-style Excalidraw diagrams for anything — architectures, pipelines, flows, papers, decisions. Renders locally to PNG, inspects its own output, and iterates. Decomposes big systems into linked component views.
Ask for a diagram. Get hand-drawn .excalidraw files that actually explain things.
Claude writes the JSON, renders it locally, looks at the image, fixes what's wrong — and only then shows you. No browser, no server, no canvas app. Just files.

One conversation → a multi-panel visual doc. Opens and stays editable on excalidraw.com.
Detail from the example above — concepts get drawn, not captioned.
As a plugin (recommended — works in every repo, auto-updates):
/plugin marketplace add sidfeels/excalidraw-diagrams
/plugin install excalidraw@excalidraw-diagrams
Or as a plain skill (this machine, all projects):
git clone https://github.com/sidfeels/excalidraw-diagrams
cp -r excalidraw-diagrams/skills/excalidraw ~/.claude/skills/
Requirements (for the local render-and-review loop):
| uv | runs the bundled renderer, zero setup |
| cairo | brew install cairo · apt-get install libcairo2-dev |
No uv/cairo? Everything still works — Claude authors valid .excalidraw files and lints their geometry; you only lose the visual self-review pass.
Just ask:
Diagrams land in excalidraw/ in your project — each .excalidraw with its rendered .png beside it (a bundled hook re-renders on every edit).
understand the subject → plan (one flow axis, one hero, ≤3 hues)
→ author terse JSON → geometric lint → render PNG
→ Claude READS the image → audit → fix → re-render
→ deliver (your chat feedback keeps iterating)
The skill ships a single helper CLI (excal.py) that handles the error-prone geometry deterministically:
| Command | Does |
|---|---|
render | .excalidraw → PNG, hand-drawn fidelity, ~1.4s |
check | lints geometry: overlaps, text overflow, dangling arrows, broken refs |
connect | edge-to-edge arrows with proper two-way bindings (straight / elbow / labeled) |
textbox | correctly centered bound text in any shape |
share | end-to-end-encrypted upload → excalidraw.com/#json=… link |
Plus ~1,000 lines of distilled diagramming methodology: layout ranking, semantic color discipline, a pattern library (fan-outs, offset stacks, timelines, evidence panels), arrow routing rules, and decomposition protocols. Read skills/excalidraw/SKILL.md — it's the product.
pycairo build fails → install system cairo first (see Requirements), then retry. Python 3.12 is pinned for wheel availability..excalidraw file is always the source of truth..excalidraw file onto excalidraw.com — everything round-trips.MIT
Modifies files
Hook triggers on file write and edit operations
Own this plugin?
Verify ownership to unlock analytics, metadata editing, and a verified badge. GitHub access is read-only (username + org membership).
Sign in to claimOwn this plugin?
Verify ownership to unlock analytics, metadata editing, and a verified badge. GitHub access is read-only (username + org membership).
Sign in to claimBased on adoption, maintenance, documentation, and repository signals. Not a security audit or endorsement.
npx claudepluginhub sidfeels/excalidraw-diagrams --plugin excalidrawPermanent coding companion for Claude Code — survives any update. MCP-based terminal pet with ASCII art, stats, reactions, and personality.
Comprehensive skill pack with 66 specialized skills for full-stack developers: 12 language experts (Python, TypeScript, Go, Rust, C++, Swift, Kotlin, C#, PHP, Java, SQL, JavaScript), 10 backend frameworks, 6 frontend/mobile, plus infrastructure, DevOps, security, and testing. Features progressive disclosure architecture for 50% faster loading.
Intelligent prompt optimization: injects the right context at the right moment so Claude lands a better first output. Clarifies vague prompts with research-based questions, plus targeted nudges for approach selection, plan readability, workflow routing, background execution, subagent routing, output readability, user-decision questions, and plan-mode assessment
Memory compression system for Claude Code - persist context across sessions
Feature development with code-architect/explorer/reviewer agents, CLAUDE.md audit and session learnings, and Agent Skills creation with eval benchmarking from Anthropic.
Claude harness - A harness for solo developers (Vibecoders) to handle full-cycle contract development.