Unify Logo Footer.svg
AI Governance
Logo
UnifyApps

UnifyApps

Logo

3 mins READ

Introduction

UnifyApps governs its own agents through the same mechanism as everyone else's: a connection, a sweep, and an explicit registration step.

There is no shortcut for being the native platform, and that is deliberate. One inventory, one policy model and one incident trail are only meaningful if the platform's own agents are in them on the same terms.

What UnifyApps discovers

A UnifyApps connection returns the agents that tenant hosts, and supports model import as well. Each agent becomes a record in the Agent Registry.

Capability support

Capability

Support

Maturity

Discovery of agents

Yes

GA

Model import

Yes

GA

Manual agent registration

Yes

GA

Trace ingestion

Yes

GA

Observe · Advisory

Yes

GA

Control · Enforced

Yes

GA

Note: You do not instrument a discovered agent. Its traces arrive through the connection — there is no SDK step and nothing to install. A discovered agent shows Setup Pending until its first trace lands, which usually means it simply has not run yet, not that you owe it work. Give it real usage and it flips to Active on its own. While it is pending, the agent detail page still shows a Setup tab carrying SDK and OTLP instructions. On a discovered agent those do not apply — the tab appears for anything awaiting a first trace, however it was registered. See Agent Statuses.

Expect a large result

The sweep returns your deployed UnifyApps agents — on a tenant that has been in use for a while, that can still be a lot.

A large sweep is normal, and it is the strongest argument for treating registration as a deliberate act rather than turning on Auto-register agents immediately.

  1. Sort by activity first — Agents with recent traces are the ones doing work. Everything else is history.

  2. Register what runs — Coverage of the active set is worth far more than a registry containing every prototype anyone ever built.

  3. Leave the rest Excluded — Visible, inventoried, ungoverned — and a defensible decision rather than an oversight.

"A three-thousand-agent sweep is not three thousand risks. It is one inventory and a triage job."

Add the connection

Step 1: Open + Add Application in Discovery

Step 2: Select UnifyApps

Step 3: Choose the connection type

Both Observe · Advisory and Control · Enforced are available for UnifyApps. Choose Control · Enforced if you want Block and Mask to take effect; on Observe · Advisory only Flag does.

Step 4: Choose the connection and click Connect

Verify

The row shows PLATFORM as UnifyApps, and DISCOVERED populates once the sweep completes — often with a large count. Open the connection and work the Assets table by activity.

Limitations

  1. Actions depend on the tier you choose — On Observe · Advisory only Flag takes effect. On Control · Enforced, Block and Mask take effect too.

  2. Volume is the real constraint — The work here is triage, not connection setup.

  3. Auto-register is a poor first choice — On a tenant with thousands of agents it converts a review queue into a registry nobody curated.

Troubleshooting

Thousands of agents appeared. Is that wrong? Almost certainly not. It is every deployed agent on the tenant. Sort by activity and register what runs.

Most of them show 0 traces. Expected — deployed agents nobody is using produce no traces. That column is your triage key.