What this is for
The Agents tab on the main Dashboard summarizes every AI agent Guardway has discovered across your connected cloud providers. It answers “how many agents do we have, how many route through our gateways, and where are the OWASP findings concentrated?” without opening the full Agents workspace. Security and platform leads use it as the daily snapshot; operators jump into Agents via Open Agents workspace when they need the inventory table or per-agent detail.Options
Tab bar
The Dashboard uses the same tab strip on every visit: Overview, Usage, Spend, Security, Agents, Adversarial. Selecting Agents loads this view (?tab=agents on the overview route).
Actions
| Element | Notes |
|---|---|
| Open Agents workspace | Outline button, top right. Navigates to the Agents sidebar page (Inventory and Findings tabs). |
Insight strip
When at least one high-signal condition is present, a banner above the stats highlights the top finding — for example Most findings: Identity & Privilege Abuse with a count and View link into the Agents → Findings tab pre-filtered to that category. When nothing notable is detected, the strip shows a neutral “no notable agent risks” message.Stat rail
Six headline metrics in a single row:| Stat | Notes |
|---|---|
| Total agents | Count across every connected provider. Subtitle shows last sync timestamp (Synced <date>), Awaiting first sync, or Not connected when no discovery integration is active. |
| Via gateway | Percentage of agents whose model endpoints match a registered Guardway gateway. Subtitle breaks down N routed · N direct · N unknown. Icon turns green at ≥ 80%, red below. Shows — when no routable denominator exists. |
| Risky MCP | Count of MCP servers flagged high risk across all agents. Subtitle: High-risk MCP servers. |
| Findings | Total open OWASP findings, all severities. Subtitle splits high-impact (critical + high) from other severities, or None detected. |
| Critical + High | Count of findings at those two severities only. Subtitle lists individual critical and high counts. Sparkline uses critical tone when greater than zero. |
| Models | Distinct model deployment names in use. Subtitle lists up to two top model names, or None in use. |
Breakdown panels
Four linkable breakdown lists below the stat rail (top six rows each unless noted):| Panel | Notes |
|---|---|
| Findings by severity | One row per severity level that has open findings (Critical, High, Medium, Low, Info). Each row links to Agents → Findings with ?severity=<level>. |
| Findings by ASI category | Rows use the full category label (e.g. ASI03 · Identity & Privilege Abuse). Each row links to Agents → Findings with ?category=ASI03. |
| Models in use | Model deployment name and agent count. Links to Agents → Inventory. |
| Agents by provider | Connector name (e.g. Azure AI Foundry, Amazon Bedrock) and agent count. Links to Agents → Inventory. |
How to configure
This tab is read-only. Connect a discovery integration under Settings → Integrations and register gateways under Gateways so routing and inventory populate.
Read the stat rail
Review Total agents, Via gateway, Findings, and Critical + High for a quick health check.
Related
- Agents — Inventory, Findings, and per-agent detail.
- Topology — graph view of gateways → providers → models → agents.
- Settings → Integrations — connect Azure AI Foundry, Amazon Bedrock, and related services.