Organizations & teams
Organize agents into teams, set a tighten-only acceptance-policy hierarchy, and watch live acceptance rates — one console for humans, one clean interface for agents.
A left-nav control plane: an overview of live acceptance, a team hierarchy, a git-like policy editor, the agent roster, members & roles, and an audit feed. Intuitive, dark, fast.
An agent enrolls into a team with a join token, fetches the effective policy it's graded against, and reports liveness — over plain HTTP or MCP. It just works.
POST /api/v1/agent/enroll { "join_token": "otji-…" }
GET /api/v1/agent/policy?modality=text
POST /api/v1/agent/heartbeat { "status": "online" }Read the agent guide →Each level can only raise the bar, never lower it. An agent is always graded against the strictest combination that applies to it.
Group agents into teams — business units, cost centers, regulatory domains. Teams nest, so a parent team's policy flows down to its children.
Author acceptance policies with full version control: a commit log, diffs between versions, one-click rollback, and fork-to-tighten. Every change is auditable.
Watch each team's ship / route / block mix, throughput, cost, and agent liveness in real time — the AgentOS control plane, scoped to your org.
Mint a join token; an agent enrolls with one call and is instantly graded against its team's effective policy. No redeploy, no config drift.
Condition a policy on concrete exemplars — a gold deck, a reference image, a spreadsheet — content-addressed and deduped, inline or by object-store pointer.
Owner, admin, member, viewer. Give people exactly the access they need; the gate and every verdict stay defensible and signed.
Stop unreviewed agent work from reaching production. Manage every team to one acceptance standard — and prove it with a signed audit trail.