Use Mermaid diagrams in DecisionsAI to explain technical topics — architecture, flows, sequences, and data models via the freestanding diagram viewer.
Scanned 9/10/2026
Install to Claude Code
npx -y skills add tensology/decisionsai --skill decisions-mermaid-diagrams --agent claude-codeInstalls into .claude/skills of the current project.
Are you the author of Decisions Mermaid Diagrams?
Add the live security badge to your README — it updates automatically with every re-scan.
[](https://www.skillsdirectory.com/skills/tensology-decisions-mermaid-diagrams)More formats (shields.io, HTML) on the badges page.
---
name: decisions-mermaid-diagrams
description: Use Mermaid diagrams in DecisionsAI to explain technical topics — architecture, flows, sequences, and data models via the freestanding diagram viewer.
---
# Decisions Mermaid Diagrams
When explaining something **technical** — architecture, request flows, state machines, database relationships, deployment topology, or multi-step processes — prefer a **Mermaid diagram** over a long prose outline.
## When to diagram
Use a diagram when the user would benefit from seeing structure or flow:
- System or module architecture
- API / request sequences
- State transitions
- ER or data relationships
- CI/CD or workflow pipelines
- Before/after refactors
Skip diagrams for trivial one-line answers, pure opinion, or non-structural content.
## Related surfaces (not Mermaid)
| Need | Skill |
|------|-------|
| Rich implementation plan with wireframes + approval gate | `visual-plan` (BuilderIO) via `decisions-visual-plan` |
| PR/branch visual recap | `visual-recap` |
| UI prototype, deck, motion, hand-drawn whiteboard | `decisions-open-design` |
Use Mermaid for **text-native** technical graphs the agent can store in diagram History. Use Open Design for **pixel/CSS artifacts** and branded deliverables.
## How to show diagrams
### In DecisionsAI chat (server agent)
**Open the viewer only** (user says "open the mermaid viewer", no diagram yet):
- Call **`open_page`** with `page="diagram viewer"`, or
- Call **`show_mermaid_diagram`** with empty `mermaid_code` (reopens the last diagram from History, or a blank editor).
Do not ask what diagram to show — open the viewer immediately.
**Show a specific diagram:**
Call **`show_mermaid_diagram`** with:
- `mermaid_code` — valid Mermaid source (no markdown fences)
- `title` — short label for the viewer window (e.g. "Auth flow", "Module layout")
This opens a **freestanding viewer window** where the user can:
- See the rendered diagram
- Edit the Mermaid source and re-render
- Export **PNG** or **JPEG**
- Copy the diagram image or source code to the clipboard
- Browse **History** in the viewer sidebar (recent diagrams are saved locally)
- **Google Drawing** export when Google is connected (Settings → Advanced)
### In chat replies
You may still include a ` ```mermaid ` fenced block in markdown. Chat renders an **Open diagram** control that opens the same viewer.
### In Cursor (this IDE)
For standalone analytical artifacts in Cursor chat, a [canvas](file:///Users/paul/.cursor/skills-cursor/canvas/SKILL.md) may still be appropriate for data-heavy output. For classic Mermaid node/graph diagrams in DecisionsAI, use `show_mermaid_diagram` or fenced mermaid blocks.
## Mermaid quality bar
- One diagram = one idea. Split overloaded charts.
- Short node labels (2–4 words). Put detail in chat prose.
- Prefer `flowchart TD` / `sequenceDiagram` / `stateDiagram-v2` / `erDiagram` as appropriate.
- Left-to-right or top-down flow; avoid spaghetti crossing edges when possible.
## Example
```text
show_mermaid_diagram(
title="Chat message path",
mermaid_code='''sequenceDiagram
participant U as User
participant W as Web UI
participant A as Agent
U->>W: send message
W->>A: POST /api/chats/{id}/messages
A-->>W: stream tokens
W-->>U: render reply'''
)
```
## Reference
- Viewer page: `/diagram/`
- API: `POST /api/diagrams` → `{ id, viewer_path }`
- Library: `/static/vendor/mermaid/mermaid.min.js` (Mermaid 11)
Is this your skill, or is something wrong with this listing? Request removal or report an issue. Author removals are honored within 72 hours.
No comments yet. Be the first to comment!