Reports

Overview

Run adversarial agent tests and repository security reports from a single Reports view.

Reports (/app/reports) lists every security test run in your organization—both agent red-team projects and repository scans—in one searchable table.

Report types

Type What it tests Typical use
Agent An AI system in production (chatbot, coding agent, voice agent, workflow, etc.) Jailbreaks, policy violations, unsafe tool use
Repository A GitHub codebase Vulnerabilities, unsafe patterns, dependency and config issues

Create either type from Reports → Create report (/app/reports/new).

Report list

The list shows name, source, type, status, and last activity. Use search and pagination to find past runs. Selecting a report opens its detail view:

  • Agent projects → /app/red-team/agents/[id] (legacy path; same project)
  • Repository reports → /app/red-team/repositories/[id] (legacy path; same report)

Findings from reports

Repository reports can produce findings that appear on the Findings page for triage. Agent projects may surface issues in the project detail UI depending on your configuration.

Prerequisites

  • Agent reports — no GitHub repo required; describe the agent and select an agent type.
  • Repository reportsSuperagent Security installed on the target repo with repository access.

Legacy navigation

The sidebar previously listed Red Teaming sub-items separately. Those flows now live under Reports:

Old area New location
Red Teaming → Agents Reports (agent type)
Red Teaming → Repositories Reports (repository type)
Safety tests Reports

Next steps