builder

A lovable.dev-style "wish an app" chat builder on Solid. You chat a wish — "build me a landing page for my honey from my bees" — a coding agent builds a real Vite + React app, and a live preview link appears in the conversation. Iterate by chatting again.

It is a thin chat UI + server-side orchestrator on top of the sibling codespaces bridge, which provides the entire build engine (repos, the issue-driven coder agent, the build runner, and Solid-Pod "Pages" publishing). The builder owns the conversation and project model; the bridge owns repos and builds. See AGENTS.md.

Shared packages (GitHub Packages)

This app installs @mind-studio/core and @mind-studio/ui from GitHub Packages. A committed .npmrc scopes @mind-studio to that registry; before installing, export a GitHub token with read:packages:

export NODE_AUTH_TOKEN=<a GitHub PAT with read:packages>
npm install

Dev setup

This prototype needs the mind-codespaces bridge running first — it does the building.

# 1. Start the shared Mind CSS (:3011), then the codespaces bridge.
(cd ../.. && docker compose up -d) # shared Mind CSS on :3011 (see ../../SOLID-SERVER.md)cd ../../codespaces
OPENROUTER_API_KEY=sk-or-...\
MIND_AGENT_MODEL=openai/gpt-oss-120b:free \
npm run dev # bridge on :3010# 2. Start the builder (points at the shared :3011 so WebIDs line up).cd ../apps/builder
cp .env.example .env.local # already targets :3010 / :3011
npm install
npm run dev # builder on :3070

Open http://localhost:3070, sign in as alice (alice@mind.local / dev-only-do-not-use-in-prod on the shared CSS), and type a wish.

Scripts

Script	Purpose
npm run dev	Next.js dev server on :3070
npm run seed:demo	Create one demo "honey landing" project end-to-end against the local bridge (idempotent)
npm run smoke:wish	Drive /api/wish, poll until published, assert the preview URL returns 200
npm run typecheck	tsc --noEmit

Ports

Service	Port
Builder / Next.js	3070
mind-codespaces bridge (required)	3010
shared Mind CSS (required)	3011

Status

MVP (M1): single wish → React/Vite SPA → preview card in chat, against a local bridge with dev-header auth. Iteration, multi-project dashboard, and the production "delegated" bridge-session path are later milestones — see the plan and AGENTS.md.

Releases

Versioning, CHANGELOG.md, and tags are automated with release-please — don't tag or edit CHANGELOG.md by hand.

1. Commit to main using Conventional Commits: fix: → patch, feat: → minor, feat!: / BREAKING CHANGE: → major. chore: / docs: / refactor: / test: don't trigger a release.
2. release-please keeps an open "chore(main): release X.Y.Z" PR that rolls the pending commits into CHANGELOG.md and bumps the version.
3. Merge that PR to release: it creates the vX.Y.Z tag + GitHub Release, which fires release.yml to build and push the Docker image to GHCR.
4. Deploying the image to production is a separate, manual GitOps step in mindpods-infra (mind-deploy.sh).