feat: calibrate continuity-aware triage risk

Author: Jacobinwwey

docs/brainstorms/2026-04-21-mainline-reconciliation-after-m8-48-triage-policy-requirements.md

@@ -5,6 +5,8 @@ topic: mainline-reconciliation-after-m8-48-triage-policy
 
 # Mainline Reconciliation After M8.48 Triage Policy Requirements
 
+> Status note (2026-04-21): `M8.49` has now shipped on `main`. This document remains the pre-implementation narrowing pass and is superseded by [Mainline Reconciliation After M8.49 Triage Prioritization Requirements](./2026-04-21-mainline-reconciliation-after-m8-49-triage-prioritization-requirements.md).
+
 ## Problem Frame
 
 `M8.48` is now shipped on `main`.
@@ -49,7 +51,7 @@ The active question is "whether bounded continuity summary fields should start a
 ## Requirements
 
 **Truth Refresh**
-- R1. This document becomes the new current source-of-truth item for post-`M8.48` direction.
+- R1. This document captured the current source-of-truth item for post-`M8.48` direction before `M8.49` landed.
 - R2. Current-facing docs must describe `M8.48` as shipped continuity-aware triage/runbook consumption.
 
 **Next-Lane Narrowing**

docs/brainstorms/2026-04-21-mainline-reconciliation-after-m8-49-triage-prioritization-requirements.md

@@ -0,0 +1,83 @@
+---
+date: 2026-04-21
+topic: mainline-reconciliation-after-m8-49-triage-prioritization
+---
+
+# Mainline Reconciliation After M8.49 Triage Prioritization Requirements
+
+## Problem Frame
+
+`M8.49` now ships on `main`.
+The bounded continuity-aware triage policy step is closed:
+
+- replay-risk severity now calibrates against persistent continuity signals,
+- operator payloads now expose `replayRiskSignals` alongside `replayRiskLevel`,
+- low-risk persistent continuity can now promote into medium risk,
+- medium replay risk with persistent continuity can now promote into high risk.
+
+That changes the next seam again.
+The active question is no longer whether continuity should affect severity policy.
+The active question is whether operator prioritization order should still stay replay-rate-only once severity has become continuity-aware.
+
+## Verified Mainline Reality (2026-04-21)
+
+- `AgentWorkspaceDiagnosticsSummary` now persists both:
+  - `replayRiskLevel`,
+  - `replayRiskSignals`.
+- `/triage` and `/triage/history` now expose continuity-aware severity buckets.
+- `topReplayReports` still sorts by raw `replayCandidateRate`, not by calibrated severity or continuity-aware priority.
+- Backend sufficiency remains green and dormant-by-default.
+
+## Critical Findings
+
+1. `M8.49` closed the severity-policy seam.
+   Persistent continuity now changes triage severity, not only runbook copy.
+
+2. Another policy-only slice would be low leverage.
+   The next mismatch is ordering, not risk classification.
+
+3. The sharpest remaining seam is prioritization.
+   A continuity-boosted high-risk report can still sit behind a raw higher replay-rate report because `topReplayReports` ordering remains replay-rate-only.
+
+4. Larger dashboards still remain secondary.
+   They would consume the same ordering later, but they do not tighten the current triage loop as directly as bounded prioritization does.
+
+## Requirements
+
+**Truth Refresh**
+- R1. This document becomes the new current source-of-truth item for post-`M8.49` direction.
+- R2. Current-facing docs must describe `M8.49` as shipped continuity-aware triage policy calibration.
+
+**Next-Lane Narrowing**
+- R3. Current-facing docs must define `M8.50` as bounded continuity-aware triage prioritization.
+- R4. `M8.50` must stay inside already-normalized fields:
+  - `replayRiskLevel`,
+  - `replayRiskSignals`,
+  - `replayCandidateRate`,
+  - `managedConversationContinuitySummary`.
+- R5. `M8.50` must not require new persistence, a broader dashboard surface, or full-report joins.
+- R6. `M8.50` must keep prioritization explainable in triage payloads rather than introducing opaque ranking.
+
+**Scope Control**
+- R7. Do not reopen backend-escalation or CI-repair narration unless fresh evidence appears.
+- R8. Do not move markdown-reader work ahead of `M8.50` unless it compounds the same operator/study loop more directly.
+- R9. Do not jump to broader operator dashboards before bounded continuity-aware prioritization lands or is explicitly rejected.
+
+## Success Criteria
+
+- this document is listed first under `docs/brainstorms/index.md`,
+- EN/ZH current-facing docs describe `M8.49` as shipped,
+- EN/ZH current-facing docs describe `M8.50` as bounded continuity-aware prioritization work,
+- docs verification passes,
+- `main` stays clean after verification.
+
+## Scope Boundaries
+
+- This document does not reopen backend sufficiency work.
+- This document does not add a new diagnostics persistence model.
+- This document does not claim that continuity-aware prioritization already ships.
+- This document does not require a new operator dashboard.
+
+## Next Steps
+
+-> `/ce:plan` for `post-M8.49 continuity-aware triage prioritization`, keeping `M8.50` bounded to explainable ordering on top of existing fast-path fields.

docs/brainstorms/index.md

@@ -4,6 +4,7 @@ This page tracks active and archived brainstorming outputs used before implement
 
 ## Current
 
+- [2026-04-21 Mainline Reconciliation After M8.49 Triage Prioritization Requirements](2026-04-21-mainline-reconciliation-after-m8-49-triage-prioritization-requirements.md)
 - [2026-04-21 Mainline Reconciliation After M8.48 Triage Policy Requirements](2026-04-21-mainline-reconciliation-after-m8-48-triage-policy-requirements.md)
 - [2026-04-21 Mainline Reconciliation After M8.47 Operator Consumer Requirements](2026-04-21-mainline-reconciliation-after-m8-47-operator-consumer-requirements.md)
 - [2026-04-21 Mainline Reconciliation After M8.46 Operator Continuity Summary Requirements](2026-04-21-mainline-reconciliation-after-m8-46-operator-continuity-summary-requirements.md)

docs/diataxis/en/explanation/agent-conversation-focus-mode-plan.md

@@ -487,12 +487,13 @@ It is conversation-product hardening on top of the shipped shell.
 
 Priority order:
 
-1. make `M8.49` about continuity-aware triage policy calibration first, so the same normalized summary affects severity policy before any broader dashboard work,
-2. expand broader operator rollups only after that policy step is in place,
-3. keep markdown-reader improvements secondary unless they clearly compound the same study loop.
+1. make `M8.50` about continuity-aware triage prioritization next, so explainable top-report ordering catches up with the continuity-aware severity policy that `M8.49` already shipped,
+2. expand broader operator rollups only after that prioritization step is in place,
+3. keep markdown-reader improvements secondary unless they clearly compound the same study/operator loop more than `M8.50`.
 
 ## Related Pages
 
+- [Mainline Reconciliation After M8.49 Triage Prioritization Requirements](../../../brainstorms/2026-04-21-mainline-reconciliation-after-m8-49-triage-prioritization-requirements.md)
 - [Mainline Reconciliation After M8.48 Triage Policy Requirements](../../../brainstorms/2026-04-21-mainline-reconciliation-after-m8-48-triage-policy-requirements.md)
 - [Mainline Reconciliation After M8.47 Operator Consumer Requirements](../../../brainstorms/2026-04-21-mainline-reconciliation-after-m8-47-operator-consumer-requirements.md)
 - [Mainline Reconciliation After M8.46 Operator Continuity Summary Requirements](../../../brainstorms/2026-04-21-mainline-reconciliation-after-m8-46-operator-continuity-summary-requirements.md)

docs/diataxis/en/explanation/development-progress-dashboard.md

@@ -6,7 +6,7 @@ It explicitly separates branch-level implementation status from mainline integra
 ## Scope
 
 - Focus area: local-first knowledge mastery platform (ingest, retrieval, learning path, tutor, memory, governance).
-- Time window: `v1.7.0` through the 2026-04-21 foundation-integration checkpoint, the 2026-04-21 UTC CI review cycle, and the post-`M8.48` reconciliation pass.
+- Time window: `v1.7.0` through the 2026-04-21 foundation-integration checkpoint, the 2026-04-21 UTC CI review cycle, and the post-`M8.49` reconciliation pass.
 - Evidence rule: every progress claim must map to:
   - contract layer (`src/learning/api.ts`, `src/learning/types.ts`),
   - route/runtime layer (`src/server.ts`),
@@ -18,13 +18,13 @@ Current top priority is not feature-count expansion.
 Current top priority is **conversation-product hardening on top of the already-green backend baseline**:
 
 - keep backend sufficiency as a standing decision gate rather than an active execution lane,
-- keep closing the correction/product/operator gap exposed after `M8.20-M8.48`,
-- treat `M8.23` as the operator-visibility baseline for conversation-memory/correction actions, `M8.24` as the active-atom action-surface baseline, `M8.25` as the active-atom pane-state continuity baseline, `M8.26` as the active-atom study-loop summary baseline, `M8.27` as the active-atom recent-result continuity baseline, `M8.28` as the active-atom result-preview baseline, `M8.29` as the active-atom history-drill-down baseline, `M8.30` as the active-atom history follow-up baseline, `M8.31` as the active-atom history freshness-ranking baseline, `M8.32` as the active-atom history rationale/alternative baseline, `M8.33` as the active-atom history confidence-signal baseline, `M8.34` as the primary-vs-secondary tradeoff explanation baseline, `M8.35` as the adjacent-history drift explanation baseline, `M8.36` as the adjacent-history stability explanation baseline, `M8.37` as the adjacent-history stability-signal baseline, `M8.38` as the managed-memory gap follow-up baseline, `M8.39` as the managed-memory continuity-transition baseline, `M8.40` as the managed-memory continuity-rollup baseline, `M8.41` as the managed-memory continuity next-step-rollup baseline, `M8.42` as the managed-memory resolved-action-rollup baseline, `M8.43` as the managed-memory latest-transition-signal baseline, `M8.44` as the managed-memory mixed latest-transition-rollup baseline, `M8.45` as the managed-memory continuity-index-rollup baseline, `M8.46` as the managed-memory continuity breadth/freshness metadata baseline, `M8.47` as the operator diagnostics continuity-summary promotion baseline, and `M8.48` as the continuity-aware triage/runbook consumption baseline,
-- treat `M8.49` continuity-aware triage policy calibration as the current highest-leverage next unit, because runbook guidance now consumes continuity context while the risk-bucket policy still ignores the same normalized digest,
-- keep broader operator dashboards and deeper personalization behind that policy-calibration seam so the next operator action can still reuse one normalized continuity digest instead of inventing another rollup path,
+- keep closing the correction/product/operator gap exposed after `M8.20-M8.49`,
+- treat `M8.23` as the operator-visibility baseline for conversation-memory/correction actions, `M8.24` as the active-atom action-surface baseline, `M8.25` as the active-atom pane-state continuity baseline, `M8.26` as the active-atom study-loop summary baseline, `M8.27` as the active-atom recent-result continuity baseline, `M8.28` as the active-atom result-preview baseline, `M8.29` as the active-atom history-drill-down baseline, `M8.30` as the active-atom history follow-up baseline, `M8.31` as the active-atom history freshness-ranking baseline, `M8.32` as the active-atom history rationale/alternative baseline, `M8.33` as the active-atom history confidence-signal baseline, `M8.34` as the primary-vs-secondary tradeoff explanation baseline, `M8.35` as the adjacent-history drift explanation baseline, `M8.36` as the adjacent-history stability explanation baseline, `M8.37` as the adjacent-history stability-signal baseline, `M8.38` as the managed-memory gap follow-up baseline, `M8.39` as the managed-memory continuity-transition baseline, `M8.40` as the managed-memory continuity-rollup baseline, `M8.41` as the managed-memory continuity next-step-rollup baseline, `M8.42` as the managed-memory resolved-action-rollup baseline, `M8.43` as the managed-memory latest-transition-signal baseline, `M8.44` as the managed-memory mixed latest-transition-rollup baseline, `M8.45` as the managed-memory continuity-index-rollup baseline, `M8.46` as the managed-memory continuity breadth/freshness metadata baseline, `M8.47` as the operator diagnostics continuity-summary promotion baseline, `M8.48` as the continuity-aware triage/runbook consumption baseline, and `M8.49` as the continuity-aware triage policy calibration baseline,
+- treat `M8.50` continuity-aware triage prioritization as the current highest-leverage next unit, because severity is now continuity-aware while top-report ordering still remains replay-rate-only,
+- keep broader operator dashboards and deeper personalization behind that prioritization seam so the next operator action can still reuse one normalized continuity digest instead of inventing another rollup path,
 - keep remote `main` CI interpretation evidence-based: the latest fully completed `main` workflow set is green, and new pushes should be treated as pending until their workflow set finishes.
 
-## Current Mainline Snapshot (2026-04-21 Baseline Split: Foundation M8.18 / Conversation M8.48)
+## Current Mainline Snapshot (2026-04-21 Baseline Split: Foundation M8.18 / Conversation M8.49)
 
 - Foundation baseline on `main` is currently:
   - status `integrated`
@@ -38,19 +38,20 @@ Current top priority is **conversation-product hardening on top of the already-g
 - Historical increment sections below remain chronological archive entries.
 - When a historical section says `in_progress` / `no-go`, read it as past checkpoint truth rather than current repo state.
 
-## Reconciliation Notes (2026-04-21 Post-M8.48)
+## Reconciliation Notes (2026-04-21 Post-M8.49)
 
 - The active mainline question is no longer whether backend sufficiency work should continue.
   That lane is closed-by-default while its gate stays green.
-- The active mainline question is also no longer whether tradeoff, drift, stability explanation, stability signaling, managed-memory continuity targeting, compact continuity rollups, compact persistent-gap next-step rollups, resolved-action retirement rollups, latest-transition signals, mixed latest-pair rollups, continuity-index rollups, bounded continuity breadth/freshness metadata, server-readable continuity-summary promotion, and continuity-aware runbook consumption should be added.
-  `M8.34`, `M8.35`, `M8.36`, `M8.37`, `M8.38`, `M8.39`, `M8.40`, `M8.41`, `M8.42`, `M8.43`, `M8.44`, `M8.45`, `M8.46`, `M8.47`, and `M8.48` now ship that continuity-explanation layer on the expanded history and operator-summary surfaces.
-- The next gap is no longer whether operators can see continuity context inside runbook actions.
-  The sharper seam is that risk-bucket policy still ignores persistent continuity signals even after runbook guidance now consumes the normalized continuity summary.
+- The active mainline question is also no longer whether tradeoff, drift, stability explanation, stability signaling, managed-memory continuity targeting, compact continuity rollups, compact persistent-gap next-step rollups, resolved-action retirement rollups, latest-transition signals, mixed latest-pair rollups, continuity-index rollups, bounded continuity breadth/freshness metadata, server-readable continuity-summary promotion, continuity-aware runbook consumption, and continuity-aware severity calibration should be added.
+  `M8.34`, `M8.35`, `M8.36`, `M8.37`, `M8.38`, `M8.39`, `M8.40`, `M8.41`, `M8.42`, `M8.43`, `M8.44`, `M8.45`, `M8.46`, `M8.47`, `M8.48`, and `M8.49` now ship that continuity-explanation layer on the expanded history and operator-summary surfaces.
+- The next gap is no longer whether persistent continuity should affect severity policy.
+  The sharper seam is that top-report prioritization still sorts by raw replay rate even after severity itself has become continuity-aware.
 - Remote `main` CI is currently green on the latest push.
   Current-facing docs must treat CI repair as dormant unless a fresh failing run exists.
 
 Execution anchor:
 
+- [Mainline Reconciliation After M8.49 Triage Prioritization Requirements (2026-04-21)](../../../brainstorms/2026-04-21-mainline-reconciliation-after-m8-49-triage-prioritization-requirements.md)
 - [Mainline Reconciliation After M8.48 Triage Policy Requirements (2026-04-21)](../../../brainstorms/2026-04-21-mainline-reconciliation-after-m8-48-triage-policy-requirements.md)
 - [Mainline Reconciliation After M8.47 Operator Consumer Requirements (2026-04-21)](../../../brainstorms/2026-04-21-mainline-reconciliation-after-m8-47-operator-consumer-requirements.md)
 - [Mainline Reconciliation After M8.46 Operator Continuity Summary Requirements (2026-04-21)](../../../brainstorms/2026-04-21-mainline-reconciliation-after-m8-46-operator-continuity-summary-requirements.md)
@@ -239,6 +240,21 @@ Execution anchor:
   - diagnostics snapshot coverage for stored preview payloads,
   - EN/ZH re-render coverage for localized preview text.
 
+## Latest Mainline Increment (2026-04-21 M8.49 Continuity-Aware Triage Policy Calibration)
+
+- Extended continuity awareness from runbook guidance into severity policy:
+  - `replayRiskLevel` now promotes one tier when normalized persistent continuity signals remain present,
+  - operator-facing metadata now persists `replayRiskSignals` so the calibrated severity remains explainable,
+  - triage/history payloads now expose those same signals next to the calibrated risk level.
+- Kept the slice bounded and explainable:
+  - continuity can only boost severity by one tier,
+  - no new persistence model, route family, or dashboard surface was introduced,
+  - `topReplayReports` ordering still remains replay-rate-only for now.
+- Expanded regression coverage for:
+  - low-to-medium continuity boosts,
+  - medium-to-high continuity boosts,
+  - continuity-aware triage/history signal payloads.
+
 ## Latest Mainline Increment (2026-04-21 M8.48 Continuity-Aware Triage/Runbook Consumption)
 
 - Extended operator-consumer logic from transport into runbook guidance:
@@ -1550,6 +1566,7 @@ Interpretation:
 
 This dashboard aligns against the following requirement chain:
 
+- [2026-04-21 Mainline Reconciliation After M8.49 Triage Prioritization Requirements](../../../brainstorms/2026-04-21-mainline-reconciliation-after-m8-49-triage-prioritization-requirements.md)
 - [2026-04-21 Mainline Reconciliation After M8.48 Triage Policy Requirements](../../../brainstorms/2026-04-21-mainline-reconciliation-after-m8-48-triage-policy-requirements.md)
 - [2026-04-21 Mainline Reconciliation After M8.47 Operator Consumer Requirements](../../../brainstorms/2026-04-21-mainline-reconciliation-after-m8-47-operator-consumer-requirements.md)
 - [2026-04-21 Mainline Reconciliation After M8.46 Operator Continuity Summary Requirements](../../../brainstorms/2026-04-21-mainline-reconciliation-after-m8-46-operator-continuity-summary-requirements.md)
@@ -1623,10 +1640,10 @@ Note: foundation-lane gates now exist on `main`; keep them in the normal verific
 
 1. Treat backend sufficiency as closed-by-default while the gate remains green; do not reopen heavier backend work without new measured pressure.
 2. Treat remote `main` CI as evidence-bound to the latest fully completed workflow set; only reopen CI-triage narrative when a fresh failing run exists.
-3. Make the next planning round about `M8.49` continuity-aware triage policy calibration:
-   - keep any new policy logic bounded to already-normalized continuity-summary fields,
-   - prefer explainable continuity-aware severity calibration before any larger operator dashboard expansion,
-   - keep markdown-reader improvements secondary unless they strengthen the same study/operator loop more than `M8.49`.
+3. Make the next planning round about `M8.50` continuity-aware triage prioritization:
+   - keep any new prioritization logic bounded to already-normalized fast-path fields,
+   - prefer explainable ordering changes before any larger operator dashboard expansion,
+   - keep markdown-reader improvements secondary unless they strengthen the same study/operator loop more than `M8.50`.
 4. Treat notification-threshold governance as closed after `M7.29`; do not plan `M7.30+` unless a materially new operator requirement appears.
 5. Keep markdown-reader governance bounded and secondary until the conversation-product lane is explicitly reprioritized.