Same agent.
Secure runtime.

Policy-governed runtime for executing AI agents and tools under explicit policy, identity, and audit controls.

Policy enforcement Tool verification Agent identity Sandboxed execution Audit trails Approval gates
Get Started GitHub
$ curl -fsSL https://symbiont.dev/install.sh | bash
01

Policy-governed execution

Every agent action is policy-gated through Cedar. Tools are cryptographically verified with SchemaPin before execution. Agents carry domain-anchored identity via AgentPin. Untrusted workloads run in Docker sandboxes. Model output is never treated as execution authority.

Explore the Trust Stack
02

Developer-first

Initialize a governed project with symbi init. Test agents with symbi run. Deploy the full runtime with symbi up. A declarative DSL for agent definitions, interactive project setup, and a built-in agent catalog.

Start building
03

Enterprise-ready

Tamper-evident audit trails for every policy decision. Slack, Teams, and Mattermost adapters. Approval gates for sensitive actions with human review. Formal Cedar authorization that your security team will actually approve.

See enterprise capabilities

Open source · Rust-native · Apache 2.0 ·