← back
AgentCraft: Putting the Orc in Orchestration — Ido Salomon
Takeaway
To scale beyond 1-2 parallel agents, treat orchestration like an RTS game: visualize the file system, batch approvals, and offload planning/review rather than per-step babysitting.
Summary
- AgentCraft is a multi-agent orchestrator that borrows RTS gaming UX: agents visualized as units on a map projection of your filesystem with rooms = files and heat maps for collision detection
- Detects/spawns live coding sessions (cursor, Claude Code, Codex, OpenCode) and lets you cycle between them with muscle memory to approve plans or answer questions
- Containers let agents run autonomous campaigns — decompose tasks, propose plans, run cron-driven idea harvesting (e.g. scrape Twitter daily and implement)
- Review bundles consolidate PRs with screenshots/video evidence so the human bottleneck shifts from babysitting to picking among 10 parallel attempts
- Workspaces enable human-to-human and human-to-agent collaboration: designers and engineers see each other's agents and can hand off work via a soft-collaboration chat
agentsorchestrationdeveloper-tools
Original description
As we run more agents in parallel, it becomes clear: we are the bottleneck. Luckily, the skills we need for effective multi-agent orchestration aren’t entirely new, they’ve just been hiding in unexpected places.Through AgentCraft, the game-inspired agent orchestrator, I’ll explore how we can raise the ceiling of human-agent collaboration without burning out in the process. Speaker info: - https://github.com/idosal - https://www.linkedin.com/in/ido-salomon/ - https://x.com/idosal1