AI Agents
The AI Agents tab under Discover AI > AI Tools lists every AI agent that sends traces to ANTS Platform. That includes coding assistants such as Claude Code, GitHub Copilot CLI and OpenAI Codex CLI, and your own autonomous agents. Agents appear automatically as they run. There is nothing to register.
Use it next to the endpoint tabs to inventory your whole AI estate: the tools people use, and the agents that run in your organization.
The tab requires the Discovered AI agents permission. Owner, Department Admin, Security & Compliance and Cost Analyst have it by default. See Members & Roles.
How agents get here
An agent appears once it sends OpenTelemetry traces to your project, for example:
- Coding agents: Claude Code, GitHub Copilot CLI and OpenAI Codex CLI, configured by the enrollment script or by hand. See Coding-Agent Telemetry.
- Your own agents: applications instrumented with the ANTS Platform SDKs or OpenTelemetry. See Integrations.
- Agents behind Portkey: agents whose traffic goes through Portkey, from its logs. See Portkey.
Agents on AWS Bedrock and Google Vertex AI are also listed from your cloud accounts by the AI Studio connectors, before they first run. Those that have not sent traces yet appear in the AI Agent Center, marked Not seen running. They join this tab once they run.
Coding agents also appear on the CLI Tools or Desktop Apps tab when the desktop agent sees them on a device. The two tabs use different sources, so their numbers differ. Neither page adds them together.
Columns
| Column | Meaning |
|---|---|
| Agent | The agent name from the traces. |
| Sanction | The agent's sanction status. An agent nobody has reviewed shows as Sanctioned, because sending traces to ANTS Platform is something your organization set up. Hover the badge to see whether the status is a default. |
| Users | Distinct people on the agent's runs, identified by an email address in the trace. A fully automated agent can have runs and zero users. |
| Sessions | Distinct sessions from people. The session ID from the trace is used when present. |
| Requests | Every run, from people or automated. |
| Tokens | Total tokens across runs. |
| Est. Cost | Estimated cost. Shows a dash when no run in the period has a price. |
| First Seen, Last Active | First and last activity in the period. |
There is no risk column. Agents have no risk level in the Tool Registry. Risk for an agent is shown in its detail panel in the AI Agent Center. The agent's full record (owner, purpose, risk tier, data sources, models, tools and permissions) is on its agent profile.
Using the tab
- Time range: Last 24 hours, Last 2 days, Last 3 days or Last 7 days (default). The page-wide filters do not apply on this tab.
- Search: filter agents by name.
- Headline figures: if you can open the AI Agent Center, the strip above the tabs shows its figures for agents (total agents, requests, cost, latency and risk). Otherwise the AI Tools figures stay in place.
- Agent detail: if you can open the AI Agent Center, select a row to open the agent's detail panel from it.
- Shortcuts: AI Agent Center and Agent Traces open those views in a new browser tab. AI Agent Center is shown only if your role can open it.
Related
- AI Command Center
- Traces & Observations
- AI Usage Intelligence: Top AI agents by users
- AI Tools, Registry & Policy