Kits / 04 · Claude Agent Starter

Claude Agent Starter

An autonomous Claude agent — tools, memory, approvals, web UI. The hard parts of an agent already wired together, so you start from a running system.

Launch price€39
Get it — €39 Back to all kits
What's inside

An agent, fully wired.

  • Tool loopthe core reason-act cycle that drives the agent's work.
  • Persistent memorystate that survives across turns and sessions.
  • Approval gatea human-in-the-loop checkpoint before risky actions run.
  • Streaminglive token output so you watch the agent think.
  • CLI + web UIdrive it from the terminal or a browser, same core.
  • MCP-readyplug in MCP servers to extend what the agent can do.
Requirements

Bring two things.

Node 20+ and your own Anthropic API key. No framework lock-in, no vendor to wire up beyond your key. You ship code, not compute.

License

Yours to keep.

Single-developer commercial license with lifetime updates. Use it in your own products. When the API changes, you get the refreshed code — no re-purchase.

Get it — €39