Agent Studio
Where governed agents are built, chat-tested, and published — behind a reliability gate.
category ▸ Agents & Identity
Agent Studio, in plain language.
An agent studio is the authoring environment for building, testing, and shipping AI agents. It is where an agent's instructions, tools, and guardrails are assembled and where a builder validates behavior before the agent reaches real work.
What distinguishes a governed studio from a generic builder is the gate on the way out: an agent does not simply go live because someone clicked publish. It must clear a quality bar, and the registry can refuse a release that does not meet it.
How Cortex implements it.
This term isn't abstract here — it maps to a real capability in the runtime. Here is exactly how Cortex enforces or relates to it.
In Cortex, the agent registry's publish step consults the live reliability score and refuses to publish an agent that has sufficient run signal and scores below the bar (409 RELIABILITY_TOO_LOW). New and low-data agents publish freely; a scoring outage fails open and never blocks ops.
An improvement queue surfaces low-scoring runs to triage, fix, and re-evaluate — closing the loop from a failing eval to a recovered score that re-opens the gate.
Keep building the vocabulary.
These terms sit next to this one in the governed-AI model — follow the thread to see how the controls connect.
Reliability score
A 0–100 quality score per agent that gates which versions are allowed to publish.
Eval suite
A set of graded test cases that score an agent's quality and gate releases against regressions.
Control Tower
The live operations console for the whole agent fleet — watch, gate, pause, or kill.
Governed runtime
The execution layer where every agent action passes a policy gate before it can run.
See Agent Studio enforced, not just defined.
Book a walkthrough and watch the controls in this glossary return real verdicts, seal real evidence, and trace every fact back to its source.