Local DAG Builder MCP
Click on "Install Server".
Wait a few minutes for the server to deploy. Once ready, it will show a "Started" state.
In the chat, type
@followed by the MCP server name and your instructions, e.g., "@Local DAG Builder MCPStart the local DAG builder"
That's it! The server will respond to your query, and you can continue using it as needed.
Here is a step-by-step guide with screenshots.
Local DAG Builder MCP
A private-by-default, local-first DAG editor launched through Model Context Protocol (MCP). Build causal diagrams visually, validate cycles immediately, and save/export portable JSON, SVG, or PNG.

flowchart LR
User --> Codex
Codex --> MCP[MCP server]
MCP --> Launcher[local process manager]
Launcher --> Web[React DAG website]
Web --> Storage[Browser storage / JSON]Quick start
git clone https://github.com/resace3/local-dag-builder-mcp.git
cd local-dag-builder-mcp
npm install
npm run build
npm run start:mcpThen add the MCP configuration below and ask Codex: Start the local DAG builder. The tool returns a verified http://127.0.0.1:8080 URL. For manual web development: npm run dev.
Codex configuration
Codex uses an mcp_servers table. Add this to ~/.codex/config.toml (Windows: C:\\Users\\<you>\\.codex\\config.toml):
[mcp_servers.local_dag_builder]
command = "node"
args = ["C:\\path\\to\\local-dag-builder-mcp\\packages\\mcp-server\\dist\\index.js"]On macOS/Linux, use /absolute/path/to/local-dag-builder-mcp/packages/mcp-server/dist/index.js. In WSL, use the Linux path. Restart Codex, then ask it to call get_dag_builder_status. See docs/mcp-tools.md for details.
Features
Drag, edit, duplicate, colour, search and auto-arrange typed nodes.
Directed React Flow edges with immediate self-loop and cycle rejection.
Live validity, warnings, counts, and topological order.
Undo/redo, local save/load, JSON import/export/copy, SVG and PNG exports.
MCP tools:
start_dag_builder,stop_dag_builder,get_dag_builder_status,create_example_dag,validate_dag,open_dag_file.
Commands
npm run build · npm run dev · npm run start:web · npm run start:mcp · npm test · npm run test:e2e · npm run lint · npm run typecheck
scripts/setup.ps1 and scripts/setup.sh install and build in one step.
DAG schema
{ version: "1.0", name, nodes: [{ id, label, type, description?, color?, position }], edges: [{ id, source, target, label? }] }. Examples live in examples/.
Security and privacy
The website binds only to 127.0.0.1; no analytics or external services are used. The MCP server validates ports, uses spawn without a shell, caps request sizes, and only reads example files or paths under the project / explicit allowed folder. DAG JSON is data, never executable code. Details: docs/architecture.md.
Testing and troubleshooting
Run npm test, npm run typecheck, then npm run test:e2e. If port 8080 is busy, request another port (or the launcher picks the next available port). Delete browser local storage from the editor’s Clear action if a saved graph needs resetting.
MIT licensed.
This server cannot be installed
Resources
Unclaimed servers have limited discoverability.
Looking for Admin?
If you are the server author, to access and configure the admin panel.
Latest Blog Posts
- Your AI Chatbot Just Exposed Your CEO's Salary to an InternBy Om-Shree-0709 on .Agent IdentityMCP SecurityOAuth Delegation
- Why MCP Servers Need Execution Sandboxing (And Why Your Current Stack Isn't Enough)By Om-Shree-0709 on .Agentic AiPrompt InjectionWebAssembly
MCP directory API
We provide all the information about MCP servers via our MCP API.
curl -X GET 'https://glama.ai/api/mcp/v1/servers/resace3/local-dag-builder-mcp'
If you have feedback or need assistance with the MCP directory API, please join our Discord server