OsaCortex control center
Run every agent inside a governed boundary.
Control runtime isolation, permissions, models, and resources from one place, without slowing down the teams building with agents.
| Name | Runtime | Version | Status | Resources |
|---|---|---|---|---|
| Research VM | ag_01KXZ | 0.1.0 | Online | 4 CPU · 8 GB |
| Writer VM | ag_02QMA | 0.1.0 | Online | 2 CPU · 4 GB |
| Secure Runner | ag_03CRT | 0.2.1 | Online | 8 CPU · 16 GB |
Four questions every Agent boundary answers.
Before work starts, and for as long as it runs.
Identity
Who can request and operate an agent.
Policy
What tools, models, and data it may use.
Runtime
Where work executes and how resources are capped.
Audit
What happened, who approved it, and when.
Governance that stays visible to operators.
Keep Agent work inside explicit boundaries and grant only what each task needs, with capacity, lifecycle, and ownership set before runtime begins.
Isolated execution
Separate agent workloads by VM, workspace, and policy boundary, with sandbox and network limits enforced by default.
Explicit permissions
Grant only the tools, data access, and model endpoints each agent needs for the task in front of it.
Resource control
Set compute, concurrency, and usage limits, and keep operational visibility over every running agent.
- IdentityRequested by Finance Ops, approved by the runtime ownerVerified
- Policy3 tools, 1 model endpoint, read-only ledger accessEnforced
- RuntimeSecure Runner, 8 CPU and 16 GB, isolatedOnline
- Audit12 events today, all approvals loggedStreaming