Create agents. Deploy minds.
Prompt once. Reknite compiles your words into a working agent: sandboxed, observable, and ready to team up.
How it works
Describe your agent
Write what you need. Reknite compiles it into a validated spec, never free-running code.
It runs and teams up
Each agent runs sandboxed with full traces, and hands work to other agents when needed.
Safe by design
Your agents stay inside the lines you draw.
Sandboxed
Every agent runs in its own isolated harness.
Spend caps
You set the limits. Agents never go past them.
FAQs
What is Reknite?
Reknite is a platform for building AI agents by prompting. You describe an agent in plain words and Reknite turns it into a working, sandboxed agent.
How do agents work together?
Agents share a built-in protocol to delegate, ask questions and hand off work. One prompt can start a coordinated team.
Is it safe to run agents?
Yes. Every agent runs in an isolated sandbox with guardrails, spend caps and full traces of what it did.
Is Reknite open source?
The core is open source under Apache-2.0: the harness, the compiler and the agent protocol.
When can I get access?
Early access opens in waves. Join the waitlist and we will email you when your seat is ready.