Full matching run
Governed shortlist, review and offer preparation beyond advisory live scores.
/admin/matchmakingFuture Spec 10 →Current-state summary
The full workspace supports subject selection, profile review, temporary criteria, a shortlist and a Send designer offers modal. What is not visible is the durable Match Run lifecycle promised by the product: run ID, queued/running state, criteria version, evaluated time, stale state, comparison, override reasons, history or rerun linkage. The current screen therefore behaves as a mutable matching form around seeded results rather than a reproducible governed run.
Interface inventory
Anatomy and verified behavior
Verification states separate observed behavior from requirements or assumptions.
| Element | Current behavior | Verification |
|---|---|---|
| Run identity | Policy v2 is shown, but no Match Run ID, timestamp, criteria hash or engine version appears. | Absent live |
| Shortlist | Counter and add/remove actions are present. | Verified live |
| Offer composer | Shortlisted candidates, summary, response deadline and pre-acceptance disclosure are present. | Verified live |
| Run history/compare | No tabs or history surface appears. | Absent live |
User journeys
Current flows
Governed run to outreach
- Validate readiness
- Create immutable run
- Evaluate cohort
- Review explanations
- Save shortlist
- Prepare offers
- Confirm disclosure and send
State model
Current and expected states
Temporary adjustment panel.
Seeded candidate cards appear without completion metadata.
Not represented.
Not represented.
Offer modal appears.
Product rules
Non-negotiable boundaries
- Completed runs are immutable and reproducible.
- Rerun creates a linked successor and never mutates prior results.
- A shortlist belongs to one run and subject.
- Outreach validates run freshness and disclosure at send time.
- Human selection and override require reasons and audit.
Known current limitations · 5 mapped
What is missing, broken or unverified
Future Spec 10 owns closure →- No immutable run identity or lifecycle is visible.
- No compare, run history, override, exclusion or candidate-error experience exists.
- Seeded results can outlive their subject selection.
- Offer creation is reachable from inconsistent state.
- Actual execution and send were not tested.
Future alignment
Required evolution
- 01Implement the Match Run lifecycle and workspace tabs from Spec 10.
- 02Add comparison, history, freshness, failure and partial-result handling.
- 03Require send-time subject/run/shortlist integrity validation.
- 04Preserve outreach and assignment linkage back to the source run.
Current baseline
Acceptance record
- Current run-related controls are documented without claiming an execution engine.
- Offer preparation is evidenced but sending is not.
- Missing immutable-run states are explicit.
- Stale-result outreach is a release blocker.
Reverse-spec completeness
Documentation coverage
The interface is still changing, so visual evidence and repository tracing remain intentionally incomplete.
Purpose and user outcome
DocumentedRoles and access
DocumentedRoutes and entry points
DocumentedPage and component anatomy
DocumentedFields and displayed data
DocumentedPrimary actions
DocumentedForms and validation
DocumentedStates and transitions
DocumentedEmpty, loading and error states
DocumentedResponsive behavior
PartialAccessibility behavior
PartialActivity and audit events
PartialData sources and persistence
DocumentedNotifications and automation
PartialKnown defects and limitations
DocumentedReusable component dependencies
DocumentedFuture-spec conflicts
PartialVisual and repository evidence
DeferredAcceptance of current baseline
Documented