Getting started
From sign-up to building from your design.
Windy gives the coding agent you already use a shared source of truth for how your software is meant to work. Here's the whole path — set Windy up, tell your agent it's there, design together, then build and run from that design.
Step 1 · Set up
Connect Windy to the agent you already use.
Sign up — your first project is ready
Signing up spins up your first project automatically, named My Project. Rename it to match what you're building — and add more projects anytime you want a separate space, one per system you'd rather keep apart.
Create an MCP endpoint
Inside your project, create a project-scoped MCP endpoint. It's revocable and token-scoped, so any agent you connect sees exactly this one project's docs — never the rest of your org.
Connect your coding agent
Point Claude Code, Codex, or any MCP-aware agent at that endpoint. The endpoint page in your dashboard gives you the exact, copy-paste config for each agent.
Step 2 · Point the agent at Windy
Tell your agent the design lives in Windy.
Add a short note to your CLAUDE.md / AGENTS.md so every session knows the design docs live in Windy — not the repo — and reaches for them before touching code.
## Design docs live in Windy (MCP)
The specs & designs for this project are in the `windy` MCP, not the repo.
Before changing code, read the relevant doc with the `mcp__windy__docs_*`
tools (start at index.md). Keep them in sync: when behaviour changes, update
the doc. For larger work, drive from a Windy plan and report progress as you
complete each task.Step 3 · Design
Design with the agent. Review as the architect.
Hand over your high-level intent
In a prompt, give the agent the shape of the system — the entities, the flows, the database and framework you're leaning toward — and ask it to write the detailed specs and design into Windy. You set the direction; it does the writing.
Review in your dashboard
Open the docs in the Windy dashboard and read what came back. Got feedback? Tell the agent what to change and let it revise — you stay the architect throughout.
Have the agent gap-check itself
Ask the agent whether the docs are detailed enough to build from. If it surfaces gaps, decide which are real, then ask it to fill them. Keep looping — review, revise — until the design is solid.
Organize so the agent can find things
As the doc set grows, sort files into folders — say backend, frontend, platform, website. There's no fixed rule; the point is simply that your agent can locate the right doc later. Use your judgment.
Step 4 · Build
Build from the design — directly, or as a plan.
Small change? Implement straight from the doc
When the work is simple, point the agent at the spec and let it build. Name the doc path from Windy in your prompt — e.g. “Read /backend/specs/checkout.md from the Windy MCP and implement the checkout feature.”
Big or long-running? Turn it into a plan
For cross-cutting work or a run that lasts hours — a migration, a new subsystem, a sweeping refactor — ask the agent to create a Plan in Windy: the work broken into small, ordered tasks.
Review the plan, then let it run
Check the plan in your dashboard and have the agent iterate on any feedback. When it's right, tell it to start executing — then watch the Windy dashboard for live progress across the whole plan and each task as it completes.
Keep going