Review Queue
Check, keep or throw out what eode learned from your models' builds and solves before it becomes shared knowledge.
Where: Review Queue
As your models are built and solved, eode notices patterns that might help future builds. Each one waits here as a candidate until someone checks it. The queue is not where model approvals happen; those are in workflow runs.
The Review Queue sits at the foot of the rail. Ctrl/⌘+K and Go to Review Queue also gets you there, and RIA can take you there from the console.
See what's queued
The header counts <n> candidates. You see the candidates learned from models you own; an org admin
sees all of them.
Filter by status: All, Detected, Verified, Promoted, Rejected, Dismissed, each with a count. All leaves out dismissed and rejected candidates. Rejected includes ones the check itself turned down.
Each row has the candidate's title and summary, then badges: the learning round it came from, how deep a review it needs (three levels, the deepest an expert gate), what kind of finding it is, its status, and MODEL DELETED when its source model is gone. A line under the badges counts how often it was seen, in how many models, and how many places it would apply. Once checked, a note says what the check found.
Check a candidate
- Click Verify. eode re-solves a test set of models with the candidate applied. This takes a while.
- Read the note: verified: with what it found, or rejected: with why.
- Decide:
| Button | Does |
|---|---|
| Promote | Makes a verified candidate shared knowledge for future builds. Needs an org admin, at every review level. |
| Reject | Reject for good — a future detect won't re-surface it. Reinstate to undo. |
| Dismiss | Remove from queue. A future detect with the same finding will mint a fresh candidate. |
| Reinstate | Put this candidate back in the queue at 'detected' so it can be reviewed again. |
| Delete | Delete this candidate permanently. No confirmation. |
Everyone with a verified row sees Promote; for anyone who isn't an org admin it's refused with promoting a candidate into shared knowledge requires an org-admin approver. Acting on a candidate whose model you don't own is refused with only the owner of the candidate's source model (or an org admin) may curate it.
Find candidates now
Run detection looks through every model's solves for new candidates straight away. You get back the ones you're allowed to see.
Auto-detect decides whether eode looks on its own: Off, On build (after a model build), On solve (after a scenario solve) or All. It's one setting for the whole organization, so only an org admin can change it. Everyone else sees the current value, with Auto-detect is an org-wide setting; only an org admin can change it.
Prune <n> orphans appears when candidates are left over from deleted models (Delete candidates
whose source model has been removed.). An orphan has no owner, so in practice only an org admin can
prune.
When the list is empty or won't load
| You see | Means |
|---|---|
| Loading candidates… | Still loading. |
| Couldn't load candidates. with the reason and Retry | The list didn't arrive. Click Retry. |
| No candidates here. | Nothing for you yet. Click Run detection. |
Related
- Roles and personas: who's an org admin
- Share and govern: model ownership