A clearer view behind every deal.
Follow the research, the accepted observations and the results that reached the dashboard. Ten categories, each with its own evidence.
Snapshot as of
Not yet available
Latest saved run for each lane.
Awaiting a successful storage read.
Publication receipt present and applied.
Latest run per lane, as of this snapshot; no agent heartbeat.
24 hours. Ten research lanes.
See outcomes from saved run records, grouped into their assigned hours. Publication and verified listings are shown separately below.
Checking the dedicated 24-hour record across all ten categories.
Applied to dashboard
A publication receipt confirms that the result was applied. Eligible listings may later expire or lose permission.
Observation retained
The observation was saved, but this result was not applied to the current dashboard snapshot.
No publication receipt
Publication remains unverified. A completed run or accepted batch does not prove the dashboard changed.
The category desk.
Current listing counts are separate from the latest accepted batch. Listing changes are shown only when the publication evidence supports a comparison.
Every run has an outcome.
Up to 100 recent run records. Category cards above use a separate latest-per-category record, so an older lane remains visible.
Reading the run ledger.
Checking the saved run lifecycle and publication receipts.
Accepted listings describe the immutable observation batch. Eligible listings describe what can currently appear after freshness and permission checks. Neither a recent request nor an open lease proves an agent is online.
What these records can tell you
Run and lease times
Assigned, started, finished and lease-until values come from saved records. Assigned time can be an hour bucket or a manual run time; it does not establish the native scheduler’s due time or lateness. Expired-run finalization can record the lease deadline as the finish time, so that interval is not measured worker runtime. This read does not contact or start a researcher.
Listing changes
Applied results are compared with their evidenced predecessor. Price or shipping changes take precedence over detail changes, so a listing is counted once. Timestamp, spot and computed-premium updates alone count as unchanged. Missing comparison evidence stays unavailable.
Snapshot and batch counts
The latest accepted batch can differ from the current dashboard snapshot. Current eligibility also checks freshness and provider permission. A missing count is shown as a dash, never an assumed zero.
Scheduling is separate
This desk shows recorded execution and publication. It does not establish whether a scheduler is enabled or an agent is currently online. Schedule configuration is reviewed separately from these request-time records.