> creating or editing architecture diagrams, flowcharts, UML, ER diagrams, network topology, org charts, mind maps, or any draw.io diagram. This skill defines the .drawio XML format, style strings, edge rules, and color palette. Consult before your first edit in a new conversation.
npx skills add https://github.com/pandazki/pneuma-skills --skill pneuma-diagram
You create and edit draw.io diagrams. The user sees a live preview that updates as you write .drawio files. Changes appear in real time during streaming — no confirmation needed.
The diagram canvas is the live surface where you and the user meet. Files (.drawio XML) are the source of truth; the viewer renders them and exposes a few channels for the user to point at things and for you to drive the canvas. Use these channels — don't ask the user to describe what they're looking at.
Before you respond, scan the latest user turn for two viewer-emitted blocks:
<viewer-context> — current canvas state. Carries the active .drawio file (workspace-relative path) and, when the user clicked a shape or edge to chat about it, the selected element (cell id, value, and style summary). Trust this over your own assumptions about what's open.<user-actions> — discrete events since the last turn (file switches, page changes, element selections). These are the breadcrumbs of what the user just did on the canvas.If the user says "this box" or "that arrow" without further context, it almost always refers to the selected element from <viewer-context> — Read that file and locate the cell by id before editing.
A ViewerAddress is the one JSON object that names "which object in the viewer" — shared by locator cards, the capture action, navigation, and selection reports. Diagram's address vocabulary:
| Key | Granularity | Meaning |
|-----|-------------|---------|
| file | coarse | Workspace-relative path to the .drawio file. |
| nodeId | fine | draw.io cell id of a single node / connector. Omit → the whole diagram. |
Selection reports it on the Address: line inside <viewer-context> — copy it verbatim back into a locator card.
After creating or substantially updating a diagram, embed a locator card so the user can jump to it from the chat. The viewer renders the card as a clickable chip that opens the diagram in the preview pane. The address attribute is a ViewerAddress JSON object.
Real example:
<viewer-locator label="Open architecture.drawio" address='{"file":"architecture.drawio"}' />
For a multi-page diagram, one card opens the whole file; the user uses the viewer's page tabs to switch between <diagram> pages.
The viewer exposes one agent-callable action via POST $PNEUMA_API/api/viewer/action:
scaffold — Reset the active diagram to empty state. clearPatterns: ["(active file)"]. No params.Use it when the user asks for a clean slate on the current diagram (e.g. "start over", "clear this diagram"). The action wipes the .drawio file's content; you then write a fresh <mxfile> skeleton.
curl -X POST "$PNEUMA_API/api/viewer/action" \
-H "Content-Type: application/json" \
-d '{"action":"scaffold","params":{}}'
After scaffold, write the new diagram with Write — don't expect the canvas to do anything until you save the file.
.drawio file can contain multiple <diagram> pages — the viewer shows tabs to switch between them. Use multiple pages for related diagrams on the same topic (e.g., overview + detail views). Each <diagram> must have a unique id and descriptive name.user-box, edge-api-db) — the user can click elements on the canvas, and selections come back via <viewer-context> keyed by id. Random ids like cell-1 make those references hard to reason about.Read the file first to preserve existing cell IDs and structure.<mxfile>
<diagram id="page-1" name="Page-1">
<mxGraphModel dx="0" dy="0" grid="1" gridSize="10"
math="0" shadow="0" adaptiveColors="auto">
<root>
<mxCell id="0" />
<mxCell id="1" parent="0" />
<!-- elements here -->
</root>
</mxGraphModel>
</diagram>
</mxfile>
id="0" (root) and id="1" (default layer, parent="0") must always be the first two cells.parent="1" (or a container/group id)."server-1", "edge-api-db".vertex="1"; edges need edge="1" — mutually exclusive.<mxGeometry relative="1" as="geometry" /> — non-negotiable.<!-- --> in output.adaptiveColors="auto" on mxGraphModel for dark mode.&, <, >, ".compressed="true".10. Preserve existing IDs when editing — changing IDs breaks connections and viewer state.
Semicolon-separated key=value pairs, ending with ;. No spaces around = or ;.
rounded=1;whiteSpace=wrap;html=1;fillColor=#dae8fc;strokeColor=#6c8ebf;
ellipse;, rhombus;, swimlane;0 or 1#RRGGBB, none, or defaultwhiteSpace=wrap;html=1; on vertices so labels wrap correctlyRounded rectangle:
<mxCell id="svc-1" value="Auth Service" style="rounded=1;whiteSpace=wrap;html=1;" vertex="1" parent="1">
<mxGeometry x="100" y="100" width="140" height="60" as="geometry"/>
</mxCell>
Diamond (decision):
<mxCell id="dec-1" value="OK?" style="rhombus;whiteSpace=wrap;html=1;" vertex="1" parent="1">
<mxGeometry x="300" y="90" width="120" height="80" as="geometry"/>
</mxCell>
Database:
<mxCell id="db-1" value="PostgreSQL" style="shape=cylinder3;whiteSpace=wrap;html=1;" vertex="1" parent="1">
<mxGeometry x="100" y="240" width="100" height="70" as="geometry"/>
</mxCell>
Arrow (always expanded form):
<mxCell id="e-1" style="edgeStyle=orthogonalEdgeStyle;rounded=1;" edge="1" source="svc-1" target="db-1" parent="1">
<mxGeometry relative="1" as="geometry"/>
</mxCell>
Labeled arrow:
<mxCell id="e-2" value="Yes" style="edgeStyle=orthogonalEdgeStyle;" edge="1" source="dec-1" target="svc-1" parent="1">
<mxGeometry relative="1" as="geometry"/>
</mxCell>
Swimlane container (children use parent=container-id):
<mxCell id="lane-1" value="User Service" style="swimlane;startSize=30;" vertex="1" parent="1">
<mxGeometry x="100" y="100" width="300" height="200" as="geometry"/>
</mxCell>
<mxCell id="api-1" value="REST API" style="rounded=1;whiteSpace=wrap;html=1;" vertex="1" parent="lane-1">
<mxGeometry x="20" y="50" width="120" height="60" as="geometry"/>
</mxCell>
Coordinated fill/stroke pairs for consistent diagrams:
| Theme | fillColor | strokeColor | Use for |
|-------|-----------|-------------|---------|
| Blue | #dae8fc | #6c8ebf | Services, APIs, primary |
| Green | #d5e8d4 | #82b366 | Success, databases, storage |
| Yellow | #fff2cc | #d6b656 | Warnings, queues, buffers |
| Orange | #ffe6cc | #d79b00 | External systems, triggers |
| Red | #f8cecc | #b85450 | Errors, critical paths |
| Purple | #e1d5e7 | #9673a6 | Users, actors, auth |
| Gray | #f5f5f5 | #666666 | Infrastructure, neutral |
edgeStyle=orthogonalEdgeStyle for right-angle connectors (most common)exitX/exitY/entryX/entryY (values 0-1) <mxCell id="e1" style="edgeStyle=orthogonalEdgeStyle;" edge="1" parent="1" source="a" target="b">
<mxGeometry relative="1" as="geometry">
<Array as="points"><mxPoint x="300" y="150"/><mxPoint x="300" y="250"/></Array>
</mxGeometry>
</mxCell>
Match all diagram labels to the user's language. If they write in Chinese, all text in the diagram should be Chinese.
For advanced topics, read these files when needed:
{SKILL_PATH}/references/xml-reference.md — XML structure, layers, metadata, dark mode, edge routing{SKILL_PATH}/references/style-reference.md — style properties, shapes, colors, HTML labels, shape library (AWS/Azure/GCP/K8s/UML/BPMN/ER)VitePress static site generator powered by Vite and Vue. Use when building documentation sites, configuring themes, or writing Markdown with Vue components.
> organizing neuroscience and biomedical data (MRI, EEG, MEG, iEEG, PET, microscopy, NIRS, motion capture, EMG, MR spectroscopy, behavioral), querying BIDS layouts, validating compliance, converting DICOM to BIDS, writing metadata sidecars, or creating BIDS derivatives.
Complete Shopify development reference covering Liquid templating, OS 2.0 themes, GraphQL APIs, Hydrogen, Functions, and performance optimization (API v2026-01). Use when working with .liquid files, building Shopify themes or apps, writing GraphQL queries for Shopify, debugging Liquid errors, creating app extensions, migrating from Scripts to Functions, or building headless storefronts. Triggers on "Shopify", "Liquid template", "Hydrogen", "Storefront API", "theme development", "Shopify Functions", "Polaris". Do NOT use for non-Shopify e-commerce platforms.
Use this skill when you need to write code for a page layout in the Next.js
One-click contribution flow for Open Design (nexu-io/open-design) — even for non-coders. Pick one of four cards (ship a Skill or Design System you made with OD; translate docs; fix a typo / write a blog; report a bug), the agent validates and opens a PR (or issue) for you. Trigger words contribute to open design, ship my OD skill, ship my OD design system, translate OD docs, report an OD bug, od-contribute.
Use when creating, editing, or generating draw.io diagram files (.drawio, .drawio.svg, .drawio.png). Covers mxGraph XML authoring, shape libraries, style strings, flowcharts, system architecture, sequence diagrams, ER diagrams, UML class diagrams, network topology, layout strategy, the hediet.vscode-drawio VS Code extension, and the full agent workflow from request to a ready-to-open file.
>- component folder structure, styled-components, hooks, named exports, barrel files, layout components, and theme usage. Use when editing any file under redisinsight/ui/**, writing or modifying React components, Redux slices, styled-components, custom hooks, or when the user mentions UI, frontend, React, Redux, or styled-components.
Convert Markdown to WeChat Official Account HTML, inspect supported providers/themes/prompts, generate article images, create drafts, write with creator styles, prepare title suggestions, and remove AI writing traces.
Take pandazki/pneuma-diagram from the repository into ~/.claude/skills for personal
use, or into .claude/skills inside a project.
The agent identifies a skill by the name field in its header. Two skills with the
same name cannot sit side by side — one of them will be ignored.