Preserve turnsUsed on crash and signal finalizers (#890)

* Preserve turnsUsed on crash and signal finalizers

Crash and signal paths previously hard-coded turnsUsed to 0 even when
mid-run snapshots had already advanced the live counter. Carry the count
on the active-run handle and sync it with every TUI and exec snapshot so
terminal writes copy the in-memory value without reading disk.

* Reseed the crash handle when rotation repoints the session

Author: TheGreatAxios

CHANGELOG.md

@@ -43,6 +43,9 @@ parallel copies under `docs/` or `scripts/notes/`. At cut time: rename
 
 ### Fixed
 
+- Crash and signal finalizers preserve `turnsUsed` from the in-memory active-run
+  handle instead of writing `0`, so a signaled or crashed session keeps the
+  turn count already persisted by mid-run snapshots.
 - Occupancy takes one dry-episode shot when the parent settles idle even if the
   live fleet 1→0 edge was never observed.
 - Dry-fleet transcript and `/status` report the outcome tally only

src/exec/runner.ts

@@ -70,7 +70,11 @@ import {
   sessionContextDir,
   sessionDir,
 } from "../session/index.js";
-import { setActiveRun } from "../session/active-run.js";
+import {
+  setActiveRun,
+  syncRunStateHandle,
+  type RunStateHandle,
+} from "../session/active-run.js";
 import {
   setActiveDisposeHost,
   clearActiveDisposeHost,
@@ -377,19 +381,35 @@ export async function runExec(config: Config): Promise<ExecResult> {
   let providerFailureObserved = false;
   let providerError: InferenceErrorLike | undefined;
   let result: ExecResult | undefined;
+  const activeRunHandle: RunStateHandle = {
+    sessionId,
+    cwd: config.cwd,
+    task,
+    startedAt,
+    turnsUsed: 0,
+    model: `${config.providerName}:${config.model}`,
+  };
 
   const persist = async (
     status: "running" | "done" | "failed" | "cancelled",
     extra?: { error?: string },
   ): Promise<void> => {
     if (finalized && status === "running") return;
     if (status !== "running") finalized = true;
+    const model = `${config.providerName}:${config.model}`;
+    const nextTurnsUsed = runSink?.getTurnCount() ?? turnsUsed;
+    syncRunStateHandle(activeRunHandle, {
+      turnsUsed: nextTurnsUsed,
+      task,
+      startedAt,
+      model,
+    });
     const snapshot = {
       status,
-      turnsUsed: runSink?.getTurnCount() ?? turnsUsed,
+      turnsUsed: nextTurnsUsed,
       task,
       startedAt,
-      model: `${config.providerName}:${config.model}`,
+      model,
       mcpServers: connectedMcp,
       ...(status !== "running" ? { finishedAt: Date.now() } : {}),
       ...(extra?.error !== undefined ? { error: extra.error } : {}),
@@ -398,13 +418,7 @@ export async function runExec(config: Config): Promise<ExecResult> {
       status === "running"
         ? saveState(config.cwd, sessionId, snapshot).then(() => {
             if (!finalized) {
-              setActiveRun({
-                sessionId,
-                cwd: config.cwd,
-                task,
-                startedAt,
-                model: `${config.providerName}:${config.model}`,
-              });
+              setActiveRun(activeRunHandle);
             }
           })
         : finalizeRunState(config.cwd, sessionId, snapshot);

src/index.ts

@@ -207,8 +207,8 @@ export async function handleFatal(
 // (e.g. a throw inside a fire-and-forget `void` call), so run.json was never
 // closed out. getActiveRun surfaces the in-flight session set by the in-flight
 // runner (TUI or exec), with
-// enough (task, startedAt, model) carried on the handle itself that no read
-// of run.json is needed — a readFile here would be exactly the kind of
+// enough (task, startedAt, model, turnsUsed) carried on the handle itself that
+// no read of run.json is needed — a readFile here would be exactly the kind of
 // unbounded crash-path I/O primeCrashReporting (src/crash/report.ts) exists
 // to avoid for git: a stalled disk or network mount would block process.exit
 // forever. The write itself goes through saveCrashState, which bypasses the
@@ -222,7 +222,7 @@ async function finalizeActiveRunOnCrash(error: unknown): Promise<void> {
   try {
     await saveCrashState(run.cwd, run.sessionId, {
       status: "crashed",
-      turnsUsed: 0,
+      turnsUsed: run.turnsUsed,
       task: run.task,
       startedAt: run.startedAt,
       finishedAt: Date.now(),
@@ -267,7 +267,7 @@ async function finalizeActiveRunOnSignal(
   try {
     await saveCrashState(run.cwd, run.sessionId, {
       status: "failed",
-      turnsUsed: 0,
+      turnsUsed: run.turnsUsed,
       task: run.task,
       startedAt: run.startedAt,
       finishedAt: Date.now(),

src/session/active-run.test.ts

@@ -0,0 +1,38 @@
+import { describe, expect, test } from "bun:test";
+
+import {
+  clearActiveRun,
+  getActiveRun,
+  setActiveRun,
+  syncRunStateHandle,
+  type RunStateHandle,
+} from "./active-run.js";
+
+describe("syncRunStateHandle", () => {
+  test("updates turnsUsed and identity fields on the live handle", () => {
+    clearActiveRun();
+    const handle: RunStateHandle = {
+      sessionId: "sess",
+      cwd: "/tmp",
+      task: "old",
+      startedAt: 1,
+      turnsUsed: 0,
+      model: "provider:old",
+    };
+    setActiveRun(handle);
+
+    syncRunStateHandle(handle, {
+      turnsUsed: 7,
+      task: "new",
+      startedAt: 42,
+      model: "provider:new",
+    });
+
+    expect(getActiveRun()).toBe(handle);
+    expect(handle.turnsUsed).toBe(7);
+    expect(handle.task).toBe("new");
+    expect(handle.startedAt).toBe(42);
+    expect(handle.model).toBe("provider:new");
+    clearActiveRun();
+  });
+});

src/session/active-run.ts

@@ -4,8 +4,8 @@
 // consulted from the crash path and from signal handlers: a run that never
 // crashes and is never signaled never has this read.
 //
-// Carries enough of the live run state (task, startedAt, model) that the
-// crash handler can build a full RunState record itself. It must not read
+// Carries enough of the live run state (task, startedAt, model, turnsUsed) that
+// the crash handler can build a full RunState record itself. It must not read
 // run.json back off disk to fill these in — an unbounded readFile on the
 // crash path has the exact failure mode primeCrashReporting (src/crash/
 // report.ts) exists to avoid for git: a stalled disk or network mount would
@@ -20,9 +20,30 @@ export interface RunStateHandle {
   cwd: string;
   task: string;
   startedAt: number;
+  turnsUsed: number;
   model?: string;
 }
 
+// Keep the crash/signal handle in step with every persisted snapshot so a
+// terminal write never falls back to turnsUsed: 0 when the live run has
+// already advanced past that.
+export function syncRunStateHandle(
+  handle: RunStateHandle,
+  snapshot: {
+    turnsUsed: number;
+    task: string;
+    startedAt: number;
+    model?: string;
+  },
+): void {
+  handle.turnsUsed = snapshot.turnsUsed;
+  handle.task = snapshot.task;
+  handle.startedAt = snapshot.startedAt;
+  if (snapshot.model !== undefined) {
+    handle.model = snapshot.model;
+  }
+}
+
 let activeRun: RunStateHandle | null = null;
 
 export function setActiveRun(handle: RunStateHandle): void {

src/session/state.test.ts

@@ -80,7 +80,7 @@ test("a straggler snapshot started before a terminal write does not overwrite it
 
 test("a persisted terminal status agrees with the active-run handle without a second call site", async () => {
   const sessionId = "sess-terminal";
-  setActiveRun({ sessionId, cwd, task: "task", startedAt: 1 });
+  setActiveRun({ sessionId, cwd, task: "task", startedAt: 1, turnsUsed: 0 });
 
   await finalizeRunState(
     cwd,

src/tui/run-snapshot-kind.test.ts

@@ -69,7 +69,13 @@ describe("a snapshot write dispatched by kind", () => {
   });
 
   test("a rotation still records the outgoing session but leaves the run crash-coverable", async () => {
-    setActiveRun({ sessionId: "old", cwd, task: "task", startedAt: 1 });
+    setActiveRun({
+      sessionId: "old",
+      cwd,
+      task: "task",
+      startedAt: 1,
+      turnsUsed: 0,
+    });
 
     await write(
       "old",
@@ -87,7 +93,13 @@ describe("a snapshot write dispatched by kind", () => {
   });
 
   test("the run-ending write records the session and disarms the handle", async () => {
-    setActiveRun({ sessionId: "last", cwd, task: "task", startedAt: 1 });
+    setActiveRun({
+      sessionId: "last",
+      cwd,
+      task: "task",
+      startedAt: 1,
+      turnsUsed: 0,
+    });
 
     await write(
       "last",

src/tui/runner/exit.test.ts

@@ -49,7 +49,7 @@ function stubQuit(args: {
       getToolCallCount: () => 0,
     },
     crashGuard: { markFinalized: () => undefined, isFinalized: () => false },
-    activeRunHandle: { task: "", startedAt: 0, model: "" },
+    activeRunHandle: { task: "", startedAt: 0, turnsUsed: 0, model: "" },
     hookManager: { dispatchPostRun: async () => undefined },
     liveSessionMode: "orchestrator",
   } as unknown as RunnerServices;

src/tui/runner/exit.ts

@@ -29,6 +29,7 @@ import {
   truncateSessionLabel,
 } from "../../session/session-label.js";
 import { clearActiveDisposeHost } from "../../session/active-host.js";
+import { syncRunStateHandle } from "../../session/active-run.js";
 import { getValidCodexToken } from "../../auth/codex/session.js";
 import { getValidXaiToken } from "../../auth/xai/session.js";
 import { suppressProviderFailurePresentation } from "../provider/failure-attempt.js";
@@ -185,12 +186,16 @@ function createRunPersistence(state: RunnerState, services: RunnerServices) {
     const model = `${state.liveSource.id}:${state.liveSource.model}`;
     // Kept in step with every persisted snapshot so the crash handler's copy
     // (activeRunHandle, read by index.ts) never lags what's actually on disk.
-    services.activeRunHandle.task = task;
-    services.activeRunHandle.startedAt = state.startedAt;
-    services.activeRunHandle.model = model;
+    const turnsUsed = services.runSink.getTurnCount();
+    syncRunStateHandle(services.activeRunHandle, {
+      turnsUsed,
+      task,
+      startedAt: state.startedAt,
+      model,
+    });
     const persisted: RunState = {
       status,
-      turnsUsed: services.runSink.getTurnCount(),
+      turnsUsed,
       task,
       startedAt: state.startedAt,
       model,
@@ -586,11 +591,25 @@ export async function createRunLifecycle(
           "session-rotation",
         );
         state.sessionId = generateSessionId();
-        // Repointed, not cleared: the process lives on, so the crash handler
-        // must keep finding this handle and close out the *new* session.
-        services.activeRunHandle.sessionId = state.sessionId;
         state.startedAt = Date.now();
         state.runTaskTitle = state.config.task;
+        const rotatedBundle = services.buildSessionSources();
+        // Repointed, not cleared: the process lives on, so the crash handler
+        // must keep finding this handle and close out the *new* session. The
+        // fields it copies reseed with the repoint — a crash inside
+        // initSessionDir/buildAgent below would otherwise stamp the outgoing
+        // session's turnsUsed (and task, startedAt, model) onto a session
+        // that has run zero turns.
+        services.activeRunHandle.sessionId = state.sessionId;
+        syncRunStateHandle(services.activeRunHandle, {
+          turnsUsed: 0,
+          task:
+            state.runTaskTitle.trim().length > 0
+              ? state.runTaskTitle.trim()
+              : "(conversation)",
+          startedAt: state.startedAt,
+          model: `${rotatedBundle.selected.id}:${rotatedBundle.selected.model}`,
+        });
         services.emitter.emit(
           "session.title",
           state.runTaskTitle.trim().length > 0
@@ -599,7 +618,6 @@ export async function createRunLifecycle(
         );
         state.workdir = sessionContextDir(state.config.cwd, state.sessionId);
         await initSessionDir(state.config.cwd, state.sessionId);
-        const rotatedBundle = services.buildSessionSources();
         state.liveSources = rotatedBundle.sources;
         state.liveDefaultSource = rotatedBundle.defaultSource;
         state.liveSource = rotatedBundle.selected;

src/tui/session-start.test.ts

@@ -1,6 +1,7 @@
 import { describe, expect, test } from "bun:test";
 
 import { withMockedModuleDuring } from "../../tests/helpers/mock-module.js";
+import { setActiveRun, clearActiveRun } from "../session/active-run.js";
 import type { RunState } from "../session/state.js";
 
 describe("createTUICrashGuard", () => {
@@ -20,6 +21,15 @@ describe("createTUICrashGuard", () => {
         },
       }),
       async () => {
+        clearActiveRun();
+        setActiveRun({
+          sessionId: "live-session",
+          cwd: "/live-cwd",
+          task: "live task",
+          startedAt: 99,
+          turnsUsed: 4,
+          model: "live-provider:live-model",
+        });
         const { createTUICrashGuard } = await import("./session-start.js");
         const guard = createTUICrashGuard(() => ({
           cwd: "/boot-cwd",
@@ -64,6 +74,8 @@ describe("createTUICrashGuard", () => {
         expect(captured[0]?.state.startedAt).toBe(99);
         expect(captured[0]?.state.error).toBe("boom");
         expect(captured[0]?.state.model).toBe("live-provider:live-model");
+        expect(captured[0]?.state.turnsUsed).toBe(4);
+        clearActiveRun();
       },
     );
   });