Watching a discovery run

Whenever you start Map code, Discover infra, or Investigate, a run card appears above the tabs on Discovery and scrolls into view automatically. Here is what it shows you:

• A title ("Code map run," "Infra discovery run," or "Investigation run"), a status badge (Queued / Running / Done / Failed / Canceled), and, once finished, a short summary (for example "3 repos · 12,400 files").
• An elapsed-time counter that keeps ticking while the run is active.
• A phase bar: a row of named steps the run passes through (code map runs go through "boot → map → finalize"; infra runs go through "Validate → Enumerate → Artifacts → Relationships → Docs"), each shown as pending, in progress (spinner), done (checkmark), or failed (X), connected by a line that fills in as you progress.
• If a repository or step fails, the specific error or cause is shown directly.
• While a code map is running, a live list of repositories being processed, each showing "Analyzing…", "Building map…", "Syncing source… X/Y files", or a final "done" state with file counts (and a "+ graph" tag if a visual code map was produced), with an overall "X / Y repos" progress bar.
• After a code map finishes, the "Generating discovery docs in the background" notice, visible for about ten minutes.
• A collapsible "Event log" (with a running event count) that lists every individual event with its sequence number: things like "Mapped acme/api — 240 files", "Syncing acme/api", "Sync 80/240 files", "Artifact → knowledge/codebase/api/graph.json", or a failure like "Failed acme/api: clone_error: ...".

You can close the run card with the "✕" once you do not need to watch it live; it does not stop the run.

Runs history

The Runs tab lists every run ever started for the project: a short run ID, its status badge, when it started, how long it took (or a live counter if still running), how many events it produced, how many code snapshots it produced, and a "View log" link that reopens that run's detail card (with its full phase bar and event log). It is exactly the same view described above, just replayed for a past run instead of a live one. If no runs exist yet, it simply says "No runs yet."

Who can do this?

Seeing runs (live or historical) only needs discovery:read. Starting a new one needs discovery:manage and a mutable project, as covered in the guide for whichever action started it.