A Tour of the App

Classifyre’s left-hand navigation is organised around one journey: connect systems, scan them, review what was found, and investigate what matters. This page walks through every screen in that order — what it shows, what you do there, and where the deeper documentation lives.

Workspaces

The screen before every other screen. Classifyre opens on the workspace directory: a card for each isolated workspace on this instance, with its name, description, any links you have pinned to it, and a live count of sources and failing sources. Cards are grouped under category headings, so a directory of sixty stays readable. Pick one to enter it — everything documented below happens inside the workspace you opened.

Use New workspace to add one, the gear on a card for its settings, categories, links and URL slug, Categories to manage the shelves themselves, and All workspaces at the top of the sidebar to come back and switch. Inside a workspace, the line under its name in the sidebar names the category it is filed under and links to everything filed alongside it.

Learn more: Workspaces — what is isolated, how addressing works, how categories and links are organised, and what governance the enterprise offering adds.

Discovery

Your at-a-glance overview. Discovery (Overview in the sidebar) is the landing dashboard, in three rows:

• The brief — total findings over the last 7, 30, or 90 days, what needs attention broken down by priority, and what has arrived today, this week and this month. Review state (open, resolved, not relevant, set aside) is a single muted line: it is bookkeeping, not the headline.
• Casework and recent scans — the cases and inquiries in flight, how many matches appeared since you last looked, and the last ten runs.
• Explore connections — a search box, your saved views, and the assets worth starting from (most critical, most lineage, biggest duplicate patterns, links people drew), each one a way into Explore.

Use it to answer “what is going on?” in ten seconds, then click any number to jump straight to the filtered detail view behind it.

Priority counts, review counts, cases and top assets are all clickable — Discovery is a launchpad, not just a report.

Explore

What an asset, a source or a glossary term is connected to, and why. Every source and every asset has a Connections tab, every glossary term opens on Where it is connected, and duplicate review has an Explore section that draws its pairs, filtered by pattern, kind, score, review state and source: a canvas of what it is connected to, by lineage, links, duplicates, shared values and meaning, each line saying why. A source’s opens on its most connected assets. Keep every connection, only those inside the source, or only those leaving it for other sources; double-click an asset to dive in, and follow the trail back out. Trace lineage upstream, downstream or both, as far as it goes; badges show what is already in a case or answers a watch. Send what you found to a case, and draw the links the data does not know about (they survive every re-scan).

The Explore card on the Overview opens the same canvas on a page of its own, starting from a search, your saved views, or a table of the assets worth starting from; a view there can be saved with a name and the layout you gave it.

Learn more: Explore: Following Connections.

Sources

The systems you’ve connected. Each source is a connection to something you already run — SharePoint, Confluence, Jira, a database, a file share, a local folder. Here you add new sources, test their connection, choose which detectors run on them, and set scan schedules. A source’s page has its findings, its assets, and its Connections: what its assets are connected to, inside the source and in others (Explore).

Learn more: Sources — including the full catalog of source types and configuration fields.

Scans

Every run, past and present. A scan is one pass over a source. This screen shows run history and live progress, counts by status (running, queued, completed, failed), and trends over time. You can also start a scan manually with the scan wizard.

Learn more: Scans — the journey of a scan, its phases, and what repeat scans do.

Assets

Everything your scans have found. Every document, page, file, or record a scan registers becomes an asset with metadata — owner, location, when it last changed. The Assets screen is the inventory: filter by new, updated, or unchanged, and see which assets carry the most findings. Each asset’s page has its findings and a Connections tab: Explore’s canvas starting from that asset, with its lineage, duplicates, shared values and concepts.

Learn more: Assets & Metadata.

Findings

The signals worth acting on. Whenever a detector spots something — a leaked secret, personal data, a tagged term, a value out of range — it raises a finding with a priority and a confidence. This screen is where you triage: filter by priority, source, or status, and resolve, ignore, or mark false positive in bulk. Your decisions stick across future scans.

Learn more: Findings & Results and Detectors.

Duplicate review

The same thing, seen everywhere. Assets that share concrete values — an email address, an ID, a name — or that read as near-duplicate text are matched into pairs and clusters.

The screen is a queue, not a graph: matches are grouped by why they matched, patterns are ranked by how much work one decision settles, and you work through pairs with the keyboard until it’s empty. What you confirm can go straight into a case or an inquiry.

Learn more: Duplicate Review and the plain-English Connections & Duplicates.

Investigations

Where findings become answers. Two tools live here: inquiries — saved questions that keep watching for matching findings — and cases — structured case files where you collect evidence, weigh hypotheses, build a timeline, and write a conclusion.

Learn more: Investigations, Inquiry, and Cases.

Glossary

Your shared vocabulary. The glossary holds the people, organisations, locations, and terms that matter to your organisation, with their aliases. It keeps humans and AI agents talking about the same entities — and when the autopilot proposes a new term or alias, you review and verify it here.

The page opens on the meaning map: every concept grouped by scheme, sized by the evidence behind it, with what turns up together. Select a concept to watch it, bring it into a case or go to its evidence; add, edit and relate concepts right on the map. The lists (concepts, entities, schemes, unbound vocabulary, proposals) are the tabs next to it.

Learn more: Glossary & Shared Vocabulary.

Harness

Mission control for the AI autopilot. The harness shows everything the autopilot does: a live activity timeline, per-run flight recordings with the reasoning behind each decision, token usage, the agents and tools available, and what the system has learned. Its Configuration tab also manages AI providers and the independent Assistant/Harness assignments, along with the Harness’s external MCP tools.

Learn more: Autopilot and its flight recorder and steering pages, plus the AI Providers guide.

Sandbox

A safe place to experiment. Sandbox runs scans in isolation, without touching your real findings — useful for trying out a new source configuration or detector before turning it on for real.

Notifications

The events worth telling you about. Failed scans, sudden spikes or drops in findings, recoveries, and first-scan completions land here (and in the bell at the top right). Filter by priority or type, mark items important, and jump straight to the affected scan or finding.

Learn more: Notifications.

Settings

Workspace-wide configuration. Software version, scan capacity, language and timezone, the MCP server that lets your own tools query Classifyre, Telegram and Slack bots, and portable workspace export/import archives.

Learn more: Settings, MCP Server.

Where to next

• Follow the data end to end in From Documents to Findings.
• See how the results get worked in Investigations.
• Ready to connect something? Start with the source catalog.