> ## Documentation Index > Fetch the complete documentation index at: https://docs.quintai.dev/llms.txt > Use this file to discover all available pages before exploring further. # Read the Overview in five minutes > What each part of the Overview page means, in what order to read it, and when a number is a gap rather than a result. **Overview** is the first page after sign-in and the one to read when you have five minutes. Read it top to bottom: coverage first, then findings, then who is at risk. ![The Overview page over the last 7 days: the status card, findings by severity and day, top risks, policy decisions and the riskiest agents](https://mintcdn.com/quint-customer-docs/E4t4pZf8IOH6MZED/images/dashboard-overview-populated.png?fit=max\&auto=format\&n=E4t4pZf8IOH6MZED\&q=85\&s=7b356a277e2aabe70c953bbd81e6e1f3) ## The scope line Under the heading, a line such as "Last 7 days · updated 10:42" tells you the window every number below covers and when it was last refreshed. The default is the last 7 days. Presets run from 6 hours to 30 days; a custom range can cover up to 30 days within the last 30. Counts are kept per whole hour in UTC, so both ends of a range round down to the hour. ## 1. Status: is the data complete? The **Status** card leads with a verdict ("No high-risk findings", "3 high-severity findings", "2 critical findings") and, beside it, a **Coverage** pill. Read the pill before the verdict. * "all 12 hosts reporting" means every enrolled machine sent telemetry in the window. The verdict stands. * "Coverage is incomplete" means some machines did not report. A quiet verdict over incomplete coverage is a gap, not an all-clear. Open the pill to see hosts reporting and hours with telemetry, then check **Data Sources**. The card's wording is deliberate about what it did not measure: | Status says | It means | | - | - | | "No high-risk findings" | Findings were counted and none was high or critical | | "No high-risk findings seen" | The window had little telemetry; absence of findings is weak evidence | | "No finding record for this window" | Nothing was recorded for this window at all | | "Findings unavailable on this API" | The count could not be read; the page is not saying zero | | "Nothing captured yet" / "Activity in this window unknown" | No activity record exists or could not be read | Below the verdict, four rows link onward: **Open incidents**, **Agents at risk**, **Undeclared agents** and **Undeclared MCP servers**. Each has a review action. Under them, **Decisions waiting for you** counts the approval requests waiting on you and when the soonest expires; it opens **Approvals**. ## 2. Findings over time A chart of findings by severity (tabs **All findings**, **Critical**, **High**, **Medium**, **Low**), with the previous window drawn as a dashed line so you can see whether this window is better or worse. **View findings** opens the **Divergence** page filtered to the same window. ## 3. Top risks Findings ranked by the risk they represent, named plainly: "Credential read, then egress", "Persistence implant", "Undeclared command", "MCP tool redefined". Each bar shows its category (Credential access, Exfiltration, Persistence, Execution, Unexpected change, Intent mismatch, Approval bypass, Tool supply chain). "No findings in the window." is a result; "Ranking unavailable on this API" is not. ## 4. Policy decisions · last 24h What rules decided in the last 24 hours, always over a fixed 24-hour window regardless of the picker. On a fleet in shadow mode this reads as recorded decisions; nothing was interrupted. See [Shadow enforcement](https://docs.quintai.dev/policies/shadow-enforcement.md). ## 5. Riskiest agents A table of agents with columns **Agent**, **Host**, **Risk**, **Why** and **Last seen**. The **Why** chips are the reason an agent ranks where it does: | Chip | Meaning | | - | - | | Undeclared | An agent nobody registered is running on the fleet | | Classification unknown | The sensor could not say what kind of agent this is | | Production | The agent runs on a host marked production | | Blind — no heartbeat | The host has stopped reporting; its risk is stale | | Degraded sensor | The host reports, but not all capture layers are live | Notes under the table ("findings not attributed", "inputs missing") say when the ranking rests on partial inputs. **All agents** opens the **Agents** page. ## What to do when a number is red * Critical or high findings: [Investigate a finding](https://docs.quintai.dev/dashboard/investigate-a-finding.md). * Agents at risk or undeclared agents: open **Agents**, filter **Unclassified**, and look at **Identification** on the agent's page. * Coverage incomplete, blind or degraded hosts: open **Data Sources**, filter **Silent** or **Degraded**, and follow [Verify and troubleshoot](https://docs.quintai.dev/install/verify-and-troubleshoot.md). * Open incidents: open **Incidents** and work the board from **Open**. > [!TIP] > > Press ⌘K (Ctrl+K) from any page to jump to a host, agent, person or session by name. This documentation is built and hosted on [Mintlify](https://mintlify.com), a developer documentation platform.