About Codepic
Codepic is a remote MCP server for AI coding agents that create editable hand-drawn diagrams in the CodePic editor. Diagrams can be revised, shared as read-only links, and exported as PNG or JSON.
Key features
- 01Create diagrams from templates
- 02Build custom diagrams with nodes and edges
- 03Inspect and update existing diagrams
- 04Edit, share, and export diagrams
Use cases
- 01Draft system architecture diagrams
- 02Create database ER diagrams
- 03Map OAuth or CI/CD flows
- 04Revise an existing diagram
MCP support
Exposes tools over remote Streamable HTTP for listing templates, creating diagrams, fetching diagram structure, and updating diagrams.
Configuration
Remote servers need no local install. Pick your client, then copy the config or install in one click.
https://codepic.cc/api/mcp/mcpclaude mcp add --transport http codepic https://codepic.cc/api/mcp/mcp{"mcpServers":{"codepic":{"type":"http","url":"https://codepic.cc/api/mcp/mcp"}}}Open in Cursor{"servers":{"codepic":{"type":"http","url":"https://codepic.cc/api/mcp/mcp"}}}Open in VS Code{"mcpServers":{"codepic":{"type":"http","url":"https://codepic.cc/api/mcp/mcp"}}}{"mcpServers":{"codepic":{"serverUrl":"https://codepic.cc/api/mcp/mcp"}}}Run commands in a terminal; paste JSON into the client’s MCP config and restart. Servers that need sign-in will prompt on first connect.
https://codepic.cc/api/mcp/mcpAuth requiredTools
6This tool takes no parameters.
Details
From the project READMECodePic MCP
Let your AI coding agent draw editable diagrams — not another Mermaid snippet you have to paste.
Generate editable hand-drawn diagrams from Cursor, Claude Desktop, and other MCP clients.
CodePic MCP lets an AI coding agent create diagrams directly in CodePic instead of returning static images, Mermaid snippets, or JSON that you have to paste manually. The result opens in the CodePic editor, where you can still move shapes, edit labels, adjust connectors, share a read-only link, and export PNG or JSON.
Quick Start
It's a remote Streamable HTTP server — no local install. First create a free API key (the cpk_... token) at https://codepic.cc/settings/api-keys (sign in → New API key → copy), then add the config for your client and replace cpk_your_api_key_here.
Cursor
Create .cursor/mcp.json in your project:
{
"mcpServers": {
"codepic": {
"url": "https://codepic.cc/api/mcp/mcp",
"headers": {
"Authorization": "Bearer cpk_your_api_key_here"
}
}
}
}
Claude Code
Add it with one command:
claude mcp add --transport http codepic https://codepic.cc/api/mcp/mcp \
--header "Authorization: Bearer cpk_your_api_key_here"
Claude Desktop
Add CodePic to your claude_desktop_config.json:
{
"mcpServers": {
"codepic": {
"type": "streamableHttp",
"url": "https://codepic.cc/api/mcp/mcp",
"headers": {
"Authorization": "Bearer cpk_your_api_key_here"
}
}
}
}
VS Code (GitHub Copilot)
Add CodePic to .vscode/mcp.json:
{
"servers": {
"codepic": {
"type": "http",
"url": "https://codepic.cc/api/mcp/mcp",
"headers": {
"Authorization": "Bearer cpk_your_api_key_here"
}
}
}
}
Codex CLI
Add CodePic to ~/.codex/config.toml:
[mcp_servers.codepic]
url = "https://codepic.cc/api/mcp/mcp"
[mcp_servers.codepic.headers]
Authorization = "Bearer cpk_your_api_key_here"
Windsurf
Add CodePic to ~/.codeium/windsurf/mcp_config.json:
{
"mcpServers": {
"codepic": {
"serverUrl": "https://codepic.cc/api/mcp/mcp",
"headers": {
"Authorization": "Bearer cpk_your_api_key_here"
}
}
}
}
Cline
Open Cline's MCP settings (cline_mcp_settings.json) and add:
{
"mcpServers": {
"codepic": {
"type": "streamableHttp",
"url": "https://codepic.cc/api/mcp/mcp",
"headers": {
"Authorization": "Bearer cpk_your_api_key_here"
}
}
}
}
Other clients
Any MCP client that supports remote Streamable HTTP works — point it at the endpoint https://codepic.cc/api/mcp/mcp with header Authorization: Bearer cpk_....
Available MCP Tools
list_templates- list available diagram templates.create_from_template- create a new diagram from a template.create_diagram- create a custom diagram with nodes and edges.get_diagram- fetch the current diagram structure bydocumentIdbefore making targeted follow-up edits.update_diagram- update an existing diagram by replacing, adding, or removing nodes and edges.
Iterate on an Existing Diagram
Use get_diagram when you already have a CodePic document and want the AI agent to inspect it before changing it. This keeps the workflow grounded in the current diagram instead of guessing from memory.
Example prompt:
Use get_diagram to inspect document <documentId>, then update the diagram by adding a Redis cache between the API server and Postgres database.
Examples
| Example | What It Generates | Prompt |
|---|---|---|
| System architecture | Next.js SaaS architecture diagram | Next.js + Supabase + Postgres + R2 + Stripe + Vercel |
| Database ER diagram | SaaS billing data model | Users, teams, subscriptions, invoices, payments |
| OAuth sequence flow | PKCE login sequence diagram | Browser, app server, identity provider, database |
| CI/CD pipeline | GitHub Actions deployment pipeline | Type-check, lint, build, preview, production |
| Data pipeline | Analytics pipeline diagram | Segment, Snowflake, dbt, Metabase |
| Iterate existing diagram | Read an existing document, then update it | get_diagram + update_diagram |
Useful Links
- CodePic: https://codepic.cc
- MCP guide: https://codepic.cc/tools/mcp
- API keys (get yours): https://codepic.cc/settings/api-keys
- ER diagram maker: https://codepic.cc/tools/er-diagram-maker
- Flowchart maker: https://codepic.cc/tools/flowchart-maker
- Wireframe tool: https://codepic.cc/tools/wireframe-tool
Community Post Angle
If you share this project, lead with the workflow:
I connected Cursor to a diagram editor through MCP, so it can generate editable architecture diagrams instead of Mermaid text.
Avoid generic product claims like "best diagram tool". Developer communities respond better to a concrete workflow and real examples.
License
MIT — see LICENSE.
Comments
New to MCP?
From what the protocol is to installing your first server, these answers cover most first questions.
An open protocol that defines how AI applications connect to external tools and data in a uniform way. Clients talk to the model; MCP servers provide tools, resources and prompt templates. They exchange standard messages, so one server can be reused by many clients.