AI sees structure, not just pixels
An image only has pixels, but Mermaid code exposes all branches and logical relationships. This allows AI to accurately understand, rewrite, complete, and extend your flowchart.
OpenMMD
Diagram and book workflows designed for the AI era
OpenMMD started as a visual Mermaid editor: people draw visually, AI receives structured Mermaid. It now also supports Book projects, where every Markdown file is a chapter, and a local MCP bridge lets AI agents write chapters into your browser and publish the whole book.
Draw intuitively without learning complex code syntax
Export PNG images to easily share with your team or add to docs
Export Mermaid code so AI can accurately understand and modify it
Export Markdown containing both a diagram preview and source code
Create Book projects from Markdown chapters and publish them as readable share pages
Use the MCP bridge so AI agents can create chapters, update books, and share them from your local browser
Studio preview
Live Mermaid rendering in the landing page
Flowchart
Map product and delivery flows
Generated Mermaid
flowchart LR
A[Idea] --> B{Ready to scope?}
B -->|Yes| C[Build diagram]
B -->|No| D[Collect context]
C --> E[Review with team]
E --> F[Ship in docs]
D --> BWhy we need Mermaid
AI struggles with image flowcharts, but excels at reading Mermaid
Humans can understand an image flowchart at a glance. But if you give AI a screenshot, it has to "guess" the structure through image recognition, often misunderstanding the relationships. Mermaid, however, is structured text code. AI reads this code like its native language, knowing exactly what every node and line means.
Preserve the human drawing experience
Even though the underlying format is code, you don't need to know any code. You can drag and connect nodes just like any modern diagram tool, leaving the tedious code generation to OpenMMD.
Local-first, secure data
Diagram data is saved locally in your browser first for speed and peace of mind. When you need cloud sync, you can also store it in your own GitHub repository.
What you can export
One workflow, serving both humans and AI
OpenMMD is designed around a simple principle: you create the flowchart once, and we provide the right format for whoever needs to read it.
Export PNG images for humans
Share a polished visual with teammates, drop it into docs, or use it in slides when people simply need to see the diagram.
Export Mermaid code for AI
Give AI the exact structured text it needs to understand your logic, generate code, or help you brainstorm edge cases.
Export Markdown for both
Generate a Markdown file that contains both a human-friendly visual preview and the AI-friendly Mermaid source code.
Who it is for
Built for people who want AI-generated code and diagrams to match reality
For product managers
Draft flows visually, export Markdown, and hand the Mermaid code to AI for improved user journeys, edge cases, and acceptance criteria.
For developers
Use Mermaid as a source file for architecture, API sequence flows, and system behavior, then iterate with AI in a format it can actually manipulate.
For indie hackers and one-person companies
Move from rough thinking to diagrams, docs, prompts, and shipping decisions in one compact workflow without introducing a heavyweight diagram stack.
For Obsidian + GitHub users
Keep a repo that stores Markdown knowledge and Mermaid artifacts together, so the same repository can power your editor workflow and your Obsidian vault sync.
Storage and sync
Local-first when you want speed, repo-based when you want portability
- Diagrams are stored locally in your browser, which keeps the product fast and keeps ownership close to you.
- You can sync data into your own GitHub repository, so the storage model fits existing developer and documentation habits.
- If you already use Obsidian with a Git-based vault, the same repository can become a shared home for diagrams, notes, and Mermaid exports.
- Markdown exports make the handoff especially clean for AI agents, repo readers, and future-you.
Book projects
Turn Markdown chapters into a shareable book
A Book project is a local-first writing workspace. Each Markdown file is a chapter, OpenMMD keeps the chapter index, and the shared page publishes a readable table of contents instead of a single-file preview.
Chapter-based Markdown
Create or upload Markdown chapters, keep them ordered, and let OpenMMD maintain the book index for sync and publishing.
Readable book sharing
Share uploads the book landing page, manifest, and every chapter to R2 under a stable book URL.
GitHub-friendly structure
The project type and book chapter index are part of the data model, so synced devices can rebuild the same book structure.
Local MCP bridge
Let AI agents write into OpenMMD without sending your browser data to an API
Install the OpenMMD browser extension and run the npm MCP server. AI agents can create books, write Markdown chapters, reorder chapters, open chapters, and publish the current book through your local browser session.
Local-first writing
The AI talks to your browser through MCP and the extension. OpenMMD still uses its own IndexedDB and project store.
Book publishing tool
`openmmd_share_book` uploads the full Markdown book and returns a public readable URL.
Agent-friendly setup
A copyable Markdown guide tells coding agents how to install the MCP server and connect to the Chrome extension.
Audience fit
Especially useful for product, engineering, and solo teams building with AI
Who tends to get the most value
Product managers who want diagrams AI can help revise instead of merely describe
Developers who need Mermaid code they can store, diff, reuse, and generate from prompts
Indie hackers and solo founders building faster with AI-assisted specs and documentation
Teams using Obsidian, GitHub, and Markdown as the source of truth for product knowledge
Give AI the diagram format it can actually work with
OpenMMD helps you keep diagrams visual for people, books structured for readers, and source text accessible to AI agents. That is the sweet spot for modern product work, development workflows, and one-person companies moving fast.