Skip to content
The control plane for AI-assisted builds

Govern everyAI-assistedproduct build

Run a real build loop against todo.md via CLI and MCP. Hosted projects, keys, billing, and support on one control plane — for Claude Code, Codex, Gemini CLI, and the hosts your team already uses.

  • Live build slots
  • todo.md synced
  • Conflict-aware claims

Install in one line

npm i -g @girardmedia/bootspring && bootspring init

Runs where your agents already work

  • Claude Code
  • Codex
  • Gemini CLI
  • Cursor
  • VS Code
  • Antigravity
  • Windsurf
  • Continue
  • OpenCode
  • Aider
  • Claude Code
  • Codex
  • Gemini CLI
  • Cursor
  • VS Code
  • Antigravity
  • Windsurf
  • Continue
  • OpenCode
  • Aider

Golden path

From idea to receipt — one governed loop

Bootspring is the control plane for AI-assisted product builds: a real task loop against todo.md, not a chatbot that forgets what it shipped.

01

Seed the contract

Turn an idea into planning truth — MASTER_PLAN, TODO.md, and a dependency-aware build queue agents can actually execute.

bootspring seed

02

Run the build loop

CLI and MCP claim the next task, write the change, run gates, and mark done with auto-commit. todo.md stays the source of truth.

bootspring build next → done

03

Ship with receipts

Dashboard surfaces projects, keys, billing, and support. Quality and handoff receipts make multi-agent work auditable — not vapor.

dashboard · quality receipts

What actually ships

A control plane, not another autocomplete.

Cursor, Copilot, and Claude Code write code. Bootspring governs the work: queue, ownership, proof, and the dashboard that keeps the team aligned.

SHIP

CLI + MCP build loop

bootspring build next / done against a dependency-aware queue. Local MCP tools for Claude Code, Codex, and hosts that speak MCP.

SHIP

todo.md as contract

Visible tasks, ownership, and handoffs. Agents and humans share one source of truth.

SHIP

Hosted control plane

Projects, API keys, billing, devices, and support on bootspring.com — the product surface operators actually use.

TRUST

Honest claim labels

Live vs preview vs not-configured stays visible. We market what ships — swarm and lab AI stay labeled until runtime proof.

PREVIEW

Agent-ready, not agent-theater

Slots, leases, and conflict scans for multiplayer builds. Parallel agents are illustrated with real mechanics — not fake green status.

SHIP

Quality gates first

Done means gates ran. Handoffs and receipts give the next agent (or human) something they can verify.