Skip to main content
The fastest way to understand red teaming is to read a finished report. Your Agents list includes sample agents with reports already run, so you can see every screen before you connect your own agent.
1

Open the Agents page

Go to Red team, Agents. Fleet health shows how many agents are Clean, Flagged, Partial, Unassessed, or Running now, and the table below lists each agent with its latest report, health, and open breaches. Agents marked Sample are examples.
2

Open a sample agent

Choose a sample agent, for example Order Support Assistant (Example). The Overview tab shows Agent health, the share of attacks that held, a status such as Needs attention, and a list of open breaches with their severity.
3

Open a breach

Choose View on a breach. You see what happened, the rule it broke, the exact attack, the agent’s reply, and the tool call it made. See Findings and evidence.
4

Read a full report

Open the Reports tab and choose a run. The report counts every result, shows what couldn’t be tested, and lists each rule with the techniques tried. See Reports.
5

Check the compliance view

Open the Compliance tab to see how the same results map to frameworks such as OWASP LLM Top 10. See Compliance view.
6

Connect your own agent

Choose Connect an agent on the Agents page, fill in the connection details, and run the connection test. See Connect your agent.

​
What good looks like

After exploring a sample, you can answer three questions.
  1. Which rules did the agent break, and how severe was each one?
  2. For one breach, what exactly was sent and what did the agent reply?
  3. What does the agent’s page say about its health, and what couldn’t be tested?

See red teaming reports in BotGauge

Open sample agents and reports right now in BotGauge.

Try BotGauge →

​
Next steps

Connect your own agent

Endpoint, headers, reply field, and the connection test.

Understand your results

The seven results, severity, and how a breach is decided.