WinDAGs Help

Start simple.

WinDAGs helps AI agents plan complicated coding work as a graph you can see, change, and run. The technical details are there when you need them.

Describe what you want to build. WinDAGs breaks it into steps, shows the dependencies, and lets you review the graph before execution.

Next-Move looks at the repo, recent files, git state, skills, and runtime setup, then suggests a path for the project.

Use the install page for Claude Code, Cursor, Codex, Gemini CLI, desktop, and any MCP-compatible client.

What should I click first?

Choose this when you know the task. Type the desired outcome in plain language, generate a DAG, inspect it, then run it if it looks right.

Choose this when the question is "what should happen next?" It starts from the project folder instead of a prompt.

Provider, model, env vars, MCP tools, safety switches, and execution context.

Prompt pieces, selected skills, structured inputs, expected outputs, context, and node results.

Skill bodies, references, examples, and what a skill contributes to a node.

Sidecar status, handoff evidence, windows, import state, and failure diagnostics.