Multi-Agent Engineering
Ship faster with spec-first multi-agent collaboration. Connect Cursor, Claude, Gemini, or any code agent with one sync command. Keep multiple agents and humans aligned across local and web, with no overwrites and clear conflict resolution.
Key Features
One-Command Sync
✓
Run sync and go
Your agent's Markdown files push and pull automatically. Works with Cursor, Claude Code, Gemini CLI, and more—no complex setup.
✓
Download and run
Get a ready-to-use sync script from setup. Choose shell, JavaScript, or Python—fits your workflow.
✓
Secure agent access
Each workspace gets its own token. Control what your agent can read and write—no shared credentials.
✓
Clean conflict flow
When you and your agent edit the same file, both versions stay. Resolve in the web app with a clear side-by-side view.
Human and Agent, Side by Side
✓
No overwrites
Concurrent edits create branches instead of overwrites. Your work and the agent's work both stay—choose how to combine them.
✓
Conflict resolution in one place
The agent pauses on conflicted paths. You resolve in the web app, then sync continues. No scattered conflict files in your repo.
✓
Know who changed what
Every version shows whether you, your agent, or the system made the change. Clear attribution for reviews and audits.
✓
Revert agent edits safely
Didn't like what the agent did? Restore a previous version. Your history is always there.
Works With Your Agent
✓
Cursor
Rules and sync setup tailored for Cursor. Your specs are ready for your agent.
✓
Claude Code
Config files generated at setup so Claude Code syncs specs out of the box.
✓
Gemini CLI
Same sync flow for Gemini. One setup, one command, any of these agents.
✓
Other agents
Use the sync script with any tool that can run a command. No vendor lock-in.
Use Cases
Co-authoring specs with code agents
Let AI agents write and update specs alongside you. Both of you work in the same docs, synced.
Keeping local and web versions in sync
Edit in your IDE or in the web app—changes flow both ways. No more "did I save that locally?"
Agent-generated documentation without overwrites
Agents can add and update docs. When you both edit the same file, neither loses work—resolve when you're ready.
Team sync with human and agent edits
Multiple people and agents working on specs. Everyone stays aligned, conflicts are clear and resolvable.