fix(FN-8498): stop legacy-adoption sweep from clearing live needs-replan signal

The KTD-8 startup adoption sweep mapped needs-replan to resume-graph, clearing
it on every engine restart. But needs-replan is not un-migrated legacy: post-U3
it is written live by the graph's plan-replan seam and is the exact status
triage's todo-rediscovery keys on to re-admit a planned todo task. Clearing it
stranded replan-loop tasks in todo forever (FN-8498 sat "ready" for 80 minutes
after a restart). Map it to preserve — the status is self-resuming as-is.

Co-Authored-By: Claude Fable 5 <noreply@anthropic.com>

.changeset/preserve-needs-replan-adoption.md

@@ -0,0 +1,7 @@
+---
+"@runfusion/fusion": patch
+---
+
+summary: Fix engine restarts stranding replan-loop tasks in To Do by clearing their needs-replan signal.
+category: fix
+dev: The KTD-8 legacy-adoption table now maps `needs-replan` to `preserve` instead of `resume-graph`; it is a live graph signal written by the plan-replan seam and consumed by triage todo-rediscovery, not un-migrated legacy state.

packages/core/src/__tests__/legacy-adoption.test.ts

@@ -121,11 +121,21 @@ describe("KTD-8 adoption table — write-site census completeness (build-failing
 
 describe("resolveLegacyStatusAdoption — every legacy (status) resumes owned", () => {
   it("triage plan-review statuses resume the graph (writers deleted in U3)", () => {
-    for (const s of ["planning", "needs-replan", "plan-review-unavailable"]) {
+    for (const s of ["planning", "plan-review-unavailable"]) {
       expect(resolveLegacyStatusAdoption(s)?.kind).toBe("resume-graph");
     }
   });
 
+  /*
+  FNXC:LegacyAdoption 2026-07-22-15:55 (FN-8498 incident):
+  needs-replan is written LIVE by the graph's plan-replan seam and is the exact key
+  triage's todo-rediscovery uses to re-admit a planned todo task. Adoption must never
+  clear it — the resume-graph mapping stranded FN-8498 in `todo` across a restart.
+  */
+  it("needs-replan is preserved — it is the graph's live replan signal, not legacy", () => {
+    expect(resolveLegacyStatusAdoption("needs-replan")?.kind).toBe("preserve");
+  });
+
   it("live human/terminal gates are preserved (never disturbed)", () => {
     for (const s of ["awaiting-approval", "failed", "error", "blocked", "done", "cancelled"]) {
       expect(resolveLegacyStatusAdoption(s)?.kind).toBe("preserve");
@@ -184,7 +194,7 @@ describe("planLegacyAdoption (U9b consumers)", () => {
   const NOW = "2026-07-19T04:40:00.000Z";
 
   it("clears every resume-graph status so the graph re-enters at its owning node", () => {
-    for (const status of ["planning", "needs-replan", "plan-review-unavailable", "queued", "triaged"]) {
+    for (const status of ["planning", "plan-review-unavailable", "queued", "triaged"]) {
       const plan = planLegacyAdoption({ status }, NOW);
       expect(plan.action, status).toBe("resume-graph");
       // Clearing the legacy status IS the re-entry: the graph owns the node again.
@@ -210,6 +220,20 @@ describe("planLegacyAdoption (U9b consumers)", () => {
     }
   });
 
+  /*
+  FNXC:LegacyAdoption 2026-07-22-15:55 (FN-8498 incident):
+  A post-cutover todo row in the plan-replan loop carries status "needs-replan" and no
+  legacyAdoptedAt stamp. The startup sweep used to clear it (resume-graph), stranding the
+  task: triage's todo-rediscovery only re-admits a planned todo task on that exact status.
+  The plan must skip it entirely — no status clear, no patch, no stamp — so the replan
+  signal survives any number of engine restarts.
+  */
+  it("survives a restart mid-replan-loop: needs-replan is skipped, never cleared (FN-8498)", () => {
+    const plan = planLegacyAdoption({ status: "needs-replan" }, NOW);
+    expect(plan.action).toBe("skip");
+    expect(plan.patch).toBeUndefined();
+  });
+
   it("backfills reviewLevel-only rows and never writes both fields", () => {
     const plan = planLegacyAdoption({ reviewLevel: 1 }, NOW);
     expect(plan.patch?.enabledWorkflowSteps).toEqual([CODE_REVIEW_GROUP_ID]);

packages/core/src/legacy-adoption.ts

@@ -11,8 +11,9 @@ cutover window fails the build instead of mass-parking rows `paused` at upgrade.
 Adoption action per legacy status (KTD-8), for the FOUNDATIONAL targets (U9 scope A):
   - resume-graph   : clear the legacy triage-owned status so the graph re-enters
                      cleanly at the owning node (planning → planning node,
-                     needs-replan → plan-replan, plan-review-unavailable →
-                     plan-review retry, queued/triaged → scheduler re-pickup).
+                     plan-review-unavailable → plan-review retry,
+                     queued/triaged → scheduler re-pickup). NOT needs-replan —
+                     that is a LIVE graph signal, preserved (see its table row).
   - preserve       : a live human/terminal gate the graph must NOT disturb
                      (awaiting-approval, awaiting-user-input, failed, error,
                      blocked, done, cancelled). Pausing is NOT a status: it is
@@ -56,7 +57,16 @@ export interface LegacyAdoptionAction {
 export const LEGACY_STATUS_ADOPTION: Readonly<Record<string, LegacyAdoptionAction>> = {
   // ── Triage plan-review statuses whose writers U3 deleted → graph re-entry ──
   "planning": { kind: "resume-graph", note: "re-enter planning node" },
-  "needs-replan": { kind: "resume-graph", note: "re-enter plan-replan node" },
+  /*
+  FNXC:LegacyAdoption 2026-07-22-15:55 (FN-8498 incident):
+  `needs-replan` is NOT legacy — post-U3 it is written live by the graph's own plan-replan
+  seam (executor.ts / scheduler.ts) and CONSUMED by triage's todo-rediscovery, which only
+  re-admits a planned todo task when status === "needs-replan". The original resume-graph
+  mapping cleared it on every engine restart, deleting the exact signal its consumer keys
+  on and stranding replan-loop tasks in `todo` forever (FN-8498 sat "ready" for 80 minutes
+  after a restart). Preserve it: the status is self-resuming — triage picks it up as-is.
+  */
+  "needs-replan": { kind: "preserve", note: "live graph replan signal — triage todo-rediscovery consumes it" },
   "plan-review-unavailable": { kind: "resume-graph", note: "plan-review retry (leased)" },
   // ── Scheduler / dispatch transient states → re-pickup ─────────────────────
   "queued": { kind: "resume-graph", note: "scheduler re-queue" },