Dashboard

The AgentKeeper dashboard gives security teams one control plane for AI agent usage across workstations, repositories, and identity groups.

Main Views

• Posture: live posture, blocked actions, warning trends, and high-risk sessions.
• AI Discovery: Discovered Findings for review decisions, Discovered Coverage for MDM and runtime proof, and Usage for activity and cost context.
• Detections: supported detector catalog, coverage by agent surface, policy response, and 30-day alert activity.
• Workstations: connected Claude Code, Claude Chat, Cursor, Codex, Gemini CLI, Google Antigravity, Windsurf, Copilot, and Cowork clients.
• Activity: searchable event stream for tool calls, policy decisions, and shield detections.
• Policies: Base Policy, policy packs, MCP restrictions, and group assignment.
• Investigations: suspicious sessions grouped into reviewable timelines.
• Settings: API keys, notification routing, telemetry setup, billing, and team management.

Event Model

Each agent event captures:

• Tool name and normalized action.
• Verdict: allowed, warned, or blocked.
• Detection pattern and severity when applicable.
• Detector IDs and policy decision trace when an alert is created.
• Session, repository, branch, host, and identity context when available.
• Policy source, including Base Policy and policy packs.

Operational Flow

1. Connect the first workstation from /setup.
2. Review the first events in /activity.
3. Tighten Runtime Shield and tool policies in /team-overview?tab=policies.
4. Add synced or custom groups when the team needs audience-specific controls.
5. Use /security and /investigations for ongoing review.