The browsable window onto everything recall knows — every tab, explained.

The dashboard

recall dashboard opens a local web app — the window onto everything recall knows. It's the same model-free read path as the CLI, made visual and navigable. Nothing leaves your machine; it's a local server reading .mind.

recall dashboard          # opens on localhost
recall tray               # same, with a tray icon / don't-close console
recall stop               # stop the server

The tabs

  • Start here — the orientation path for a fresh session (same as `recall

explain`): the load-bearing files, the must-know decisions, what's in flight, where the team burns time. Open this first in an unfamiliar repo.

  • Overview — the dashboard home: recent activity, "recently learned" knowledge,

the connection/health state, and a live activity console streaming reads (brief/recall/explain/stamp) across the CLI, MCP and dashboard as they happen.

  • Brain — the knowledge feed: every lesson/decision, newest first, rendered with

its "why", its SHA, and a "what this file touches" badge. Click a lesson to walk it as a causal-chain story — decision → consequence → the pre-edit briefing of the file it names → what breaks. Every file path is clickable.

  • Tasks — the standing intent: open tasks with their affects files and a

progress bar for checklist items. Flags tasks left open too long, or marked done with unchecked items (the contradiction the tool exists to kill).

  • Drift — the honesty light: claim-bearing notes whose file has moved on since

they were stamped (🟢 fresh / 🟡 changed / 🟠 uncommitted edits). Drift is flagged, never auto-healed — you resolve it with the owner's OK (Governance & drift).

  • Search — ask the index by concept. Renders the tracks side by side: code (by

importance), knowledge (by relevance), blast radius. Every row is clickable — code/blast open the file, knowledge opens the brain entry.

  • Code — the code map + file tree: symbols, importance, the dependency edges.

The structural half of what brief reads.

  • Graph — the dependency / causal graph: the typed edges drawn as a navigable

map, so the blast radius and the decision chains are something you can see.

  • Changelog — the same release notes the website /changelog shows, served

from the engine's own changelog (one source, no drift).

  • About — what recall is, the version, the links (GitHub, Discord), and the

"built with" credits.

The connection pill + live console

The header shows a single connection · n/4 pill folding the health of the read modes; hover for the exact, manual fix command per mode (it never auto-fixes). The floating activity console streams every read across all three surfaces (CLI, MCP, dashboard) over one shared .mind — proof, in real time, that the memory is being used and costs nothing to read.

Why a dashboard at all

The CLI and MCP are for acting; the dashboard is for seeing. It turns the abstract ("the code remembers") into something walkable — the brain as a story, the blast radius as a graph, drift as a light. It's also the fastest way to show a new teammate (or yourself, after a break) what the project actually knows.