# Implementation Proof: Quote lifecycle investigator and missed quote follow-up

Status: open
Opened: 2026-06-12

## Hypothesis
If the operator dashboard turns the quote lifecycle table into a durable quote investigator, an operator can pause or filter the live quote list, select an `Awaiting executor` or otherwise missed quote, and continue following that exact quote through command, executor result, relay failure, outcome, or unresolved state even after the rolling live table moves on.

## Problem Evidence
- The current Quote lifecycle page is a rolling live table of recent rows.
- Pausing freezes the displayed table snapshot, including stale `Awaiting executor` evidence.
- When live updates resume, the quote can leave the recent window and become hard to find again.
- The existing expandable row explains the lifecycle only while the row remains in the rendered recent set.
- This prevents practical investigation of missed quotes such as commands that later become `quote_not_found_or_finished`, `verifier_salt_unavailable`, stale command rejects, or successful submissions.

## Scope
- Improve the existing `Quote lifecycle` dashboard view, not a new dashboard product area.
- Add a durable quote lookup path by `quote_id`, `decision_id`, or `command_id` using existing history tables.
- Add a selected or pinned quote investigator that remains visible independently from the rolling table rows.
- Make pause freeze table ordering only; the selected quote investigation must still refresh from durable evidence.
- Add search or lookup controls so a quote can be found again after it leaves the recent table.
- Add filters for investigation states that matter to missed quotes:
  - awaiting executor
  - approved but command missing
  - blocked before submit
  - submission failed
  - quote not found or finished
  - strategy rejected
  - submitted or awaiting outcome
- Improve the selected quote detail so `Awaiting executor` clearly shows:
  - command id and command timestamp
  - elapsed time since command
  - whether an executor result is now stored
  - relay result or block reason if present
  - next expected stage when still unresolved
- Preserve old executor result and lifecycle rows as readable when newer timing or result fields are missing.
- Keep the implementation read-side and dashboard-only unless a small shared lifecycle normalizer change is required.

## Non-goals
- Do not change strategy selection, pricing, edge thresholds, notional limits, inventory checks, arming, signer identity, or pair enablement.
- Do not skip quotes or alter executor/relay response behavior.
- Do not add active pair, edge, notional, latency, response-policy, or retention environment variables.
- Do not add a new retention policy or new long-term analytics store in this turn.
- Do not build the full remaining analytics workbench: quote volume windows, size-bucket distributions, oracle-deviation views, and matched-only PnL analytics remain backlog work.
- Do not claim submitted quote responses are completed trades unless durable outcome and settlement evidence support that claim.
- Do not use manual kubectl rollout, image patching, or ad hoc production repair.

## Definition of Done
- The Quote lifecycle page has a stable selected or pinned quote investigator separate from the rolling table rows.
- Selecting an `Awaiting executor` row lets the operator continue following that quote after the table is paused, resumed, filtered, or updated.
- A selected quote can be refreshed or auto-refreshed from durable evidence even when it is no longer present in `recent_lifecycle_rows`.
- Lookup by `quote_id`, `decision_id`, or `command_id` returns the same normalized lifecycle shape used by the table, including stage timestamps, reason text, timing, command, execution result, and outcome fields when present.
- Problem-state filters make missed quote rows easy to isolate without hiding the selected quote investigation.
- Rows with missing old fields render as unavailable instead of crashing or disappearing.
- Bug regression coverage proves:
  - a selected quote remains inspectable after it leaves the recent live list
  - an `Awaiting executor` row can later refresh into a terminal executor or relay result
  - lookup by quote, decision, and command identifiers uses durable evidence and does not rely only on the current client-side recent rows
- Targeted dashboard tests pass.
- Full `npm test` passes.
- The operator dashboard bundle builds.
- The planning and implementation commits include `Proof:`, `Assumptions:`, and `Still fake:`.
- If tests pass, deploy only through the repo push workflow and collect live evidence:
  - repo-owned deployments are on the new image
  - Quote lifecycle still receives live rows
  - at least one durable quote lookup returns a lifecycle row
  - a missed or pending quote can be inspected by identifier after it is not relied on as a visible recent row

## Failure Conditions
- Pause still freezes the only available evidence for a selected quote.
- A selected quote cannot be found once it leaves the rolling recent table.
- The lookup path searches only the browser's current rows instead of durable history.
- The UI improves labels but still cannot answer what happened after `Awaiting executor`.
- Missed-quote filters hide the selected investigation or make old rows unreadable.
- The implementation changes trading behavior or strategy economics.
- No regression test covers the row-follow-up bug.

## Real Versus Fake
- Real in scope: dashboard investigation backed by durable quote, decision, command, executor result, and outcome rows already stored by repo-owned services.
- Still fake or incomplete: venue-native final fill truth, fee-aware realized PnL, broader quote analytics workbench, and historical aggregate drilldowns beyond the existing retained lifecycle detail and rollups.