Sectr — Deploy your agent in one click
Integrate instantly: Any agent framework, any frontend
Keep writing agents the way you already do. Just register it with Sectr, and your agent is ready for cloud deployment. Wire it up to your frontend with AG-UI.
Test locally, then cloud deploy
Deploy in one click: instant cloud-deployed agent
Develop locally with `sectr dev`, testing streaming and approvals identically to prod. When you are ready to ship it, `sectr deploy` gives it a live endpoint. No infrastructure to debug.
- Live agent preview UI. Develop your agent with a live development chat UI out-of-the-box. Shift-left your dev and iteration cycles.
- Same engine, local and cloud.`sectr dev` runs your agent on your machine the way the cloud does: sessions, approvals and streaming behave the same.
- One command to ship.`sectr deploy` builds your agent in the cloud, and instantly gives it an AG-UI endpoint.
- Secrets stay encrypted. Add your model API keys as encrypted secrets in the dashboard.
Built-in AG-UI endpoint
Point your CopilotKit frontend at Sectr. Your chat UI is done.
Every agent you deploy is also an AG-UI endpoint. Point CopilotKit's React components at your agent's URL, and streaming text, tool calls and approvals render as-is. No SDK to learn, no events to parse.
- Survives a refresh. The turn runs on Sectr, not in the browser. Reload mid-answer and the chat picks up right where it was.
- Your server stays stateless. Conversations live on Sectr, each visible only to the user who owns it. Host your app anywhere.
`$ sectr init my-agent --frontend copilotkit`
Suspend, scale to zero, resume
Serverless human-in-the-loop
When an agent has to wait, do you keep your compute sitting idle, or do you build complex infra to capture and persist its state, and resume instantly later? With Sectr you don't have to chose: Sectr provides native serverless human-in-the-loop support to your agent, no code needed.
- Scale to zero while it waits. The agent suspends at the call and nothing runs until the answer arrives, however long that takes.
- Resumes exactly where it stopped. Sectr resumes your agent with its authoritative state. The agent picks up with the full conversation, tool calls included.
- Your agent code never knows. Declarative specification of approval policies at the platform level. No checkpointer, no queue, and no scheduler, no wake-up logic in your agent code.
- Native in your chat UI. Approvals reach CopilotKit as native AG-UI interrupts, so your frontend code works seamlessly.
Fast & persisted streaming
Every event: observable and debuggable
Every token, tool call and approval streams to your users live and lands in the session journal. Replay any conversation in the Sectr dashboard to see exactly what your agent did.
- More than chat text. Tool calls and results, approval decisions and errors are journaled alongside the messages, in order.
- Understand your agent. Full event history allows you to understand why your agent did what it did.
- Same inspector locally.`sectr dev` opens your local sessions in the same dashboard you use for deployed agents.
Ready to deploy your agent?
Sectr is free during beta. Sign up with GitHub, Google, Vercel or your email and deploy your agent to the cloud in one click.