Agent Radar
Real-time graph of every agent, tool, and MCP server in your fleet — seen through four security lenses. Traffic. Policy. Risk. Shadow. Your competitors show you logs. Agent Radar shows you the shape.
Agents
3
Servers
3
Req / 1h
2,841
Blocked
1
Four lenses. One graph.
Switch between interpretive views without leaving the page. Each lens reveals a different dimension of your agent fleet — all on the same live topology.
Traffic Lens
Real-time request flow. Every hop. No blind spots.
Live edges between agents, tools, and MCP servers — weighted by request volume over the last hour. Hot paths light up automatically. Block patterns surface before they become incidents.
- Edges weighted by request volume, refreshed every 10 s
- Click any node to drill into agent detail or server config
- Blocked request count on each edge — red means action needed
- Available on all plans
Policy Lens
Starter+Overlay your rules on the live graph.
See exactly which tools are allowed, denied, or shadow-monitored for every agent–server connection. Spots uncovered paths — connections with no policy rule — before they become an audit finding.
- Color-coded edges: allowed (green), denied (red), shadow (amber)
- Highlight uncovered connections with zero policy coverage
- Click any edge to jump to the matching policy rule
- Available on Starter and above
Risk Lens
Pro+Agent risk scores painted on the topology.
Every agent node glows with its current risk tier. Critical-tier agents (90–100) appear red with a warning halo. Elevated agents pulse amber. Security teams see the highest-risk connections without leaving the graph.
- Node color maps to risk tier: green / amber / orange / red
- Risk score shown inline — no need to drill into agent detail
- High-risk traffic edges highlighted for immediate triage
- Available on Pro and above
Shadow Lens
Pro+See what policy watches but does not yet enforce.
Shows which agent–server connections are in shadow mode — policy is evaluated but violations are not blocked. Identify coverage gaps and plan your enforcement rollout with confidence.
- Shadow edges shown in a distinct dashed style
- Live counts of shadow-denied events per agent
- Readiness gauge: % of connections ready for enforcement
- Available on Pro and above
How Agent Radar works
- 1
Proxy intercepts every tool call
The ShieldAgent proxy sits between every AI agent and its MCP servers. Every tool call passes through it — no agent code changes required.
- 2
Metrics written to the audit trail
For each call: agent ID, tool name, server ID, outcome (allowed/blocked/shadow), latency, and risk signals. Written in under 1 ms, never blocks the call path.
- 3
Graph built from live metrics
Agent Radar aggregates the last hour of audit events into a weighted graph — nodes are agents and servers, edges are traffic paths. Updates every 10 seconds.
- 4
Lenses apply interpretive overlays
Select a lens to overlay additional data: policy rules (Policy Lens), risk scores (Risk Lens), or shadow coverage (Shadow Lens). The graph layout stays the same — only the signal changes.
The only live agent topology in the market
Every competitor gives you logs. Agent Radar gives you the map.
Log dashboards — search through millions of events after the fact
Agent Radar — the shape of your fleet, live, before incidents escalate
Static architecture diagrams — stale the moment they're exported
Live graph — edges appear and disappear as agents connect and disconnect
One risk score per agent — no context on which connections drive it
Risk Lens — score on every node, risky traffic paths highlighted
Policy rules as a flat list — hard to see what's covered or missing
Policy Lens — rules overlaid on the topology, uncovered paths highlighted
Ready to see your agent fleet live?
Agent Radar is available on every ShieldAgent plan. Traffic Lens is free. Policy, Risk, and Shadow Lenses unlock with Starter and Pro.