gsd-pi 2.80.0-dev.2848d9009 → 2.80.0-dev.3065f7bfa

package/src/resources/extensions/gsd/tests/auto-loop.test.ts

@@ -10,10 +10,13 @@ import {
   _resetPendingResolve,
   _hasPendingResolveForTest,
   _setActiveSession,
+  _setSessionSwitchInFlight,
+  _consumePendingSwitchCancellation,
   isSessionSwitchInFlight,
 } from "../auto/resolve.js";
 import { runUnit } from "../auto/run-unit.js";
 import { autoLoop } from "../auto/loop.js";
+import { runDispatch } from "../auto/phases.js";
 import { detectStuck } from "../auto/detect-stuck.js";
 import type { UnitResult, AgentEndEvent } from "../auto/types.js";
 import type { LoopDeps } from "../auto/loop-deps.js";
@@ -64,7 +67,7 @@ function makeMockSession(opts?: {
     verbose: false,
     basePath: process.cwd(),
     cmdCtx: {
-      newSession: (options?: { abortSignal?: AbortSignal }) => {
+      newSession: (options?: { abortSignal?: AbortSignal; workspaceRoot?: string }) => {
         opts?.onNewSessionStart?.(session);
         if (opts?.newSessionThrows) {
           return Promise.reject(new Error(opts.newSessionThrows));
@@ -76,7 +79,7 @@ function makeMockSession(opts?: {
             setTimeout(() => {
               // Simulate AgentSession.newSession() checking abortSignal after
               // its internal async work (abort()) completes — this is where the
-              // real code captures process.cwd() and rebuilds the tool runtime.
+              // real code selects a workspace root and rebuilds the tool runtime.
               // If the signal is aborted, the real code discards the session.
               opts?.onSignalCheck?.(options?.abortSignal?.aborted ?? false);
               opts?.onNewSessionSettle?.(session);
@@ -206,6 +209,28 @@ test("runUnit returns cancelled when session creation fails", async () => {
   assert.equal(pi.calls.length, 0);
 });
 
+test("runUnit clears queued switch cancellation when session creation fails", async () => {
+  _resetPendingResolve();
+
+  const ctx = makeMockCtx();
+  const pi = makeMockPi();
+  const s = makeMockSession({
+    newSessionThrows: "connection refused",
+    onNewSessionStart: () => {
+      resolveAgentEndCancelled({
+        message: "Claude Code process aborted by user",
+        category: "aborted",
+        isTransient: false,
+      });
+    },
+  });
+
+  const result = await runUnit(ctx, pi, s, "task", "T01", "prompt");
+
+  assert.equal(result.status, "cancelled");
+  assert.equal(_consumePendingSwitchCancellation(), null);
+});
+
 test("runUnit returns cancelled when session creation times out", async () => {
   _resetPendingResolve();
 
@@ -221,6 +246,34 @@ test("runUnit returns cancelled when session creation times out", async () => {
   assert.equal(pi.calls.length, 0);
 });
 
+test("runUnit consumes a cancellation queued during session switch before dispatch", async () => {
+  _resetPendingResolve();
+
+  const ctx = makeMockCtx();
+  const pi = makeMockPi();
+  let cancellationQueued = false;
+  const s = makeMockSession({
+    newSessionDelayMs: 10,
+    onNewSessionStart: () => {
+      setTimeout(() => {
+        cancellationQueued = !resolveAgentEndCancelled({
+          message: "Claude Code process aborted by user",
+          category: "aborted",
+          isTransient: false,
+        });
+      }, 0);
+    },
+  });
+
+  const result = await runUnit(ctx, pi, s, "plan-slice", "M009/S01", "prompt");
+
+  assert.equal(cancellationQueued, true);
+  assert.equal(result.status, "cancelled");
+  assert.equal(result.errorContext?.category, "aborted");
+  assert.equal(result.errorContext?.message, "Claude Code process aborted by user");
+  assert.equal(pi.calls.length, 0, "queued switch cancellation must prevent prompt dispatch");
+});
+
 test("runUnit keeps the session-switch guard across a late newSession settlement", async () => {
   _resetPendingResolve();
   mock.timers.enable();
@@ -496,10 +549,9 @@ test("runUnit proceeds when isProviderRequestReady throws (defensive) (#4555)",
   assert.equal(pi.calls.length, 0);
 });
 
-test("late-resolving newSession() after timeout receives aborted signal so tool runtime is not configured with root cwd (#3731)", async () => {
-  // When newSession() times out in runUnit(), auto-mode restores cwd to project
-  // root. If newSession() later resolves, it must NOT use process.cwd() to
-  // configure the tool runtime (which would give it root cwd, not worktree cwd).
+test("late-resolving newSession() after timeout receives aborted signal so tool runtime is not configured with stale workspace root (#3731)", async () => {
+  // When newSession() times out in runUnit(), a late resolution must not
+  // configure the tool runtime against a stale workspace root.
   //
   // The fix: runUnit creates an AbortController, aborts it on timeout, and passes
   // the signal to newSession(). AgentSession.newSession() checks the signal after
@@ -514,8 +566,8 @@ test("late-resolving newSession() after timeout receives aborted signal so tool
 
     // newSession mock simulates AgentSession.newSession() behavior:
     // after an internal delay (representing await this.abort()), it checks the
-    // abortSignal — that's where the real code would capture process.cwd() and
-    // call _buildRuntime. If aborted, the real code must discard the session.
+    // abortSignal before selecting the workspace root and calling _buildRuntime.
+    // If aborted, the real code must discard the session.
     const s = makeMockSession({
       newSessionDelayMs: 200_000, // longer than NEW_SESSION_TIMEOUT_MS (120s)
       onSignalCheck: (aborted) => {
@@ -548,7 +600,7 @@ test("late-resolving newSession() after timeout receives aborted signal so tool
       abortedWhenLateSessionSettled,
       true,
       "runUnit must pass an aborted AbortSignal to newSession() when it resolves after the session-creation timeout (#3731). " +
-      "Without this, AgentSession.newSession() captures root process.cwd() and rebuilds the tool runtime with wrong cwd.",
+      "Without this, AgentSession.newSession() can rebuild the tool runtime with a stale workspace root.",
     );
   } finally {
     mock.timers.reset();
@@ -637,7 +689,11 @@ function makeMockDeps(
     resolveMilestoneFile: () => null,
     reconcileMergeState: () => "clean",
     preflightCleanRoot: () => ({ stashPushed: false, summary: "" }),
-    postflightPopStash: () => {},
+    postflightPopStash: () => ({
+      restored: true,
+      needsManualRecovery: false,
+      message: "restored",
+    }),
     getLedger: () => null,
     getProjectTotals: () => ({ cost: 0 }),
     formatCost: (c: number) => `$${c.toFixed(2)}`,
@@ -805,6 +861,145 @@ test("autoLoop exits on terminal complete state", async (t) => {
   );
 });
 
+test("autoLoop stops before success notification when postflight stash restore needs recovery", async () => {
+  _resetPendingResolve();
+
+  const notifications: Array<{ msg: string; level: string }> = [];
+  const ctx = makeMockCtx();
+  ctx.ui.setStatus = () => {};
+  ctx.ui.notify = (msg: string, level: string) => {
+    notifications.push({ msg, level });
+  };
+  const pi = makeMockPi();
+  const s = makeLoopSession();
+  let stopReason = "";
+
+  const deps = makeMockDeps({
+    deriveState: async () => {
+      deps.callLog.push("deriveState");
+      return {
+        phase: "complete",
+        activeMilestone: { id: "M001", title: "Test", status: "complete" },
+        activeSlice: null,
+        activeTask: null,
+        registry: [{ id: "M001", status: "complete" }],
+        blockers: [],
+      } as any;
+    },
+    preflightCleanRoot: () => ({
+      stashPushed: true,
+      stashMarker: "gsd-preflight-stash:M001:test",
+      summary: "stashed",
+    }),
+    postflightPopStash: () => ({
+      restored: false,
+      needsManualRecovery: true,
+      message: "git stash pop stash@{0} failed after merge of milestone M001",
+      stashRef: "stash@{0}",
+    }),
+    sendDesktopNotification: () => {
+      deps.callLog.push("sendDesktopNotification");
+    },
+    logCmuxEvent: () => {
+      deps.callLog.push("logCmuxEvent");
+    },
+    stopAuto: async (_ctx, _pi, reason) => {
+      deps.callLog.push("stopAuto");
+      stopReason = reason ?? "";
+    },
+  });
+
+  await autoLoop(ctx, pi, s, deps);
+
+  assert.equal(stopReason, "Post-merge stash restore failed for milestone M001");
+  assert.ok(
+    notifications.some(
+      (n) => n.level === "error" && n.msg.includes("Post-merge stash restore failed for milestone M001"),
+    ),
+    "failed postflight restore must be surfaced as an error",
+  );
+  assert.ok(
+    !deps.callLog.includes("sendDesktopNotification"),
+    "must not emit milestone success desktop notification after stash restore failure",
+  );
+  assert.ok(
+    !deps.callLog.includes("logCmuxEvent"),
+    "must not emit milestone success cmux event after stash restore failure",
+  );
+});
+
+test("autoLoop marks transition merge complete before postflight recovery stop", async () => {
+  _resetPendingResolve();
+
+  const ctx = makeMockCtx();
+  ctx.ui.setStatus = () => {};
+  ctx.ui.notify = () => {};
+  const pi = makeMockPi();
+  const s = makeLoopSession();
+  let mergeCalls = 0;
+  let stopReason = "";
+
+  const deps = makeMockDeps({
+    deriveState: async () => {
+      deps.callLog.push("deriveState");
+      return {
+        phase: "executing",
+        activeMilestone: { id: "M002", title: "Next", status: "active" },
+        activeSlice: null,
+        activeTask: null,
+        registry: [
+          { id: "M001", title: "Done", status: "complete" },
+          { id: "M002", title: "Next", status: "active" },
+        ],
+        blockers: [],
+      } as any;
+    },
+    preflightCleanRoot: () => ({
+      stashPushed: true,
+      stashMarker: "gsd-preflight-stash:M001:test",
+      summary: "stashed",
+    }),
+    postflightPopStash: () => ({
+      restored: false,
+      needsManualRecovery: true,
+      message: "git stash pop stash@{0} failed after merge of milestone M001",
+      stashRef: "stash@{0}",
+    }),
+    resolver: {
+      get workPath() {
+        return "/tmp/project";
+      },
+      get projectRoot() {
+        return "/tmp/project";
+      },
+      get lockPath() {
+        return "/tmp/project";
+      },
+      enterMilestone: () => {
+        assert.fail("must not enter the next milestone after postflight recovery fails");
+      },
+      exitMilestone: () => {},
+      mergeAndExit: () => {
+        mergeCalls += 1;
+      },
+      mergeAndEnterNext: () => {},
+    } as any,
+    stopAuto: async (_ctx, _pi, reason) => {
+      deps.callLog.push("stopAuto");
+      stopReason = reason ?? "";
+      if (!s.milestoneMergedInPhases) {
+        deps.resolver.mergeAndExit("M001", ctx.ui);
+      }
+    },
+  });
+
+  await autoLoop(ctx, pi, s, deps);
+
+  assert.equal(stopReason, "Post-merge stash restore failed for milestone M001");
+  assert.equal(s.milestoneMergedInPhases, true);
+  assert.equal(mergeCalls, 1, "postflight recovery stop must not re-run an already completed transition merge");
+});
+
 test("autoLoop pauses when provider readiness cancels before dispatch", async () => {
   _resetPendingResolve();
 
@@ -2089,6 +2284,25 @@ test("resolveAgentEndCancelled without args produces no errorContext field", asy
   assert.equal(resolved.errorContext, undefined, "errorContext must not be present when no args passed");
 });
 
+test("resolveAgentEndCancelled queues cancellation that arrives during session switch", () => {
+  _resetPendingResolve();
+
+  _setSessionSwitchInFlight(true);
+  const resolved = resolveAgentEndCancelled({
+    message: "Claude Code process aborted by user",
+    category: "aborted",
+    isTransient: false,
+  });
+
+  assert.equal(resolved, false);
+  const pending = _consumePendingSwitchCancellation();
+  assert.ok(pending?.errorContext, "queued cancellation should preserve errorContext");
+  assert.equal(pending.errorContext.category, "aborted");
+  assert.equal(pending.errorContext.message, "Claude Code process aborted by user");
+  assert.equal(_consumePendingSwitchCancellation(), null);
+  _resetPendingResolve();
+});
+
 // ─── #1571: artifact verification retry ──────────────────────────────────────
 
 test("autoLoop re-iterates when postUnitPreVerification returns retry (#1571)", async () => {
@@ -2509,6 +2723,199 @@ test("autoLoop stops when worktree has no .git for execute-task (#1833)", async
   );
 });
 
+test("dispatch health check wins before stuck detection for execute-task without .git", async () => {
+  _resetPendingResolve();
+
+  const ctx = makeMockCtx();
+  const pi = makeMockPi();
+  const notifications: string[] = [];
+  ctx.ui.notify = (msg: string) => { notifications.push(msg); };
+
+  const s = makeLoopSession({ basePath: "/tmp/broken-worktree" });
+  const deps = makeMockDeps({
+    existsSync: (p: string) => !p.endsWith(".git"),
+  });
+  const result = await runDispatch(
+    {
+      ctx,
+      pi,
+      s,
+      deps,
+      prefs: undefined,
+      iteration: 1,
+      flowId: "test-flow",
+      nextSeq: () => 1,
+    },
+    {
+      state: {
+        phase: "executing",
+        activeMilestone: { id: "M001", title: "Test", status: "active" },
+        activeSlice: { id: "S01", title: "Slice 1" },
+        activeTask: { id: "T01" },
+        registry: [{ id: "M001", status: "active" }],
+        blockers: [],
+      } as any,
+      mid: "M001",
+      midTitle: "Test",
+    },
+    {
+      recentUnits: [
+        { key: "execute-task/M001/S01/T01" },
+        { key: "execute-task/M001/S01/T01" },
+      ],
+      stuckRecoveryAttempts: 1,
+      consecutiveFinalizeTimeouts: 0,
+    },
+  );
+
+  assert.equal(result.action, "break");
+  assert.equal(result.reason, "worktree-invalid");
+  assert.ok(deps.callLog.includes("stopAuto"), "should stop through worktree health check");
+  assert.ok(
+    notifications.some((n) => n.includes("Worktree health check failed") && n.includes("no .git")),
+    "should notify about missing .git",
+  );
+  assert.ok(
+    !notifications.some((n) => n.includes("Stuck on execute-task")),
+    "stuck-loop message must not mask the worktree health failure",
+  );
+});
+
+test("pre-dispatch skip resolves before dispatch health and stuck accounting", async () => {
+  _resetPendingResolve();
+
+  const ctx = makeMockCtx();
+  const pi = makeMockPi();
+  const notifications: string[] = [];
+  ctx.ui.notify = (msg: string) => { notifications.push(msg); };
+
+  const s = makeLoopSession({ basePath: "/tmp/broken-worktree" });
+  const deps = makeMockDeps({
+    existsSync: (p: string) => !p.endsWith(".git"),
+    runPreDispatchHooks: () => ({ firedHooks: ["skip-execute"], action: "skip" }),
+  });
+  const loopState = {
+    recentUnits: [
+      { key: "execute-task/M001/S01/T01" },
+      { key: "execute-task/M001/S01/T01" },
+    ],
+    stuckRecoveryAttempts: 1,
+    consecutiveFinalizeTimeouts: 0,
+  };
+
+  const result = await runDispatch(
+    {
+      ctx,
+      pi,
+      s,
+      deps,
+      prefs: undefined,
+      iteration: 1,
+      flowId: "test-flow",
+      nextSeq: () => 1,
+    },
+    {
+      state: {
+        phase: "executing",
+        activeMilestone: { id: "M001", title: "Test", status: "active" },
+        activeSlice: { id: "S01", title: "Slice 1" },
+        activeTask: { id: "T01" },
+        registry: [{ id: "M001", status: "active" }],
+        blockers: [],
+      } as any,
+      mid: "M001",
+      midTitle: "Test",
+    },
+    loopState,
+  );
+
+  assert.equal(result.action, "continue");
+  assert.ok(!deps.callLog.includes("stopAuto"), "skip hook should not stop on worktree health");
+  assert.equal(loopState.recentUnits.length, 2, "skip hook should not update stuck accounting");
+  assert.ok(
+    notifications.some((n) => n.includes("Skipping execute-task M001/S01/T01")),
+    "should notify about the skip hook",
+  );
+  assert.ok(
+    !notifications.some((n) => n.includes("Worktree health check failed") || n.includes("Stuck on execute-task")),
+    "health and stuck notifications must not run before skip hook resolution",
+  );
+});
+
+test("pre-dispatch replace resolves final unit before dispatch health and stuck accounting", async () => {
+  _resetPendingResolve();
+
+  const ctx = makeMockCtx();
+  const pi = makeMockPi();
+  const notifications: string[] = [];
+  ctx.ui.notify = (msg: string) => { notifications.push(msg); };
+
+  const s = makeLoopSession({ basePath: "/tmp/broken-worktree" });
+  const deps = makeMockDeps({
+    existsSync: (p: string) => !p.endsWith(".git"),
+    runPreDispatchHooks: () => ({
+      firedHooks: ["review"],
+      action: "replace",
+      unitType: "hook/review",
+      prompt: "review before executing",
+      model: "review-model",
+    }),
+  });
+  const loopState = {
+    recentUnits: [
+      { key: "execute-task/M001/S01/T01" },
+      { key: "execute-task/M001/S01/T01" },
+    ],
+    stuckRecoveryAttempts: 1,
+    consecutiveFinalizeTimeouts: 0,
+  };
+
+  const result = await runDispatch(
+    {
+      ctx,
+      pi,
+      s,
+      deps,
+      prefs: undefined,
+      iteration: 1,
+      flowId: "test-flow",
+      nextSeq: () => 1,
+    },
+    {
+      state: {
+        phase: "executing",
+        activeMilestone: { id: "M001", title: "Test", status: "active" },
+        activeSlice: { id: "S01", title: "Slice 1" },
+        activeTask: { id: "T01" },
+        registry: [{ id: "M001", status: "active" }],
+        blockers: [],
+      } as any,
+      mid: "M001",
+      midTitle: "Test",
+    },
+    loopState,
+  );
+
+  assert.equal(result.action, "next");
+  assert.equal(result.data?.unitType, "hook/review");
+  assert.equal(result.data?.finalPrompt, "review before executing");
+  assert.equal(result.data?.hookModelOverride, "review-model");
+  assert.ok(!deps.callLog.includes("stopAuto"), "replace hook should not stop on execute-task health");
+  assert.deepEqual(
+    loopState.recentUnits.map((u) => u.key),
+    [
+      "execute-task/M001/S01/T01",
+      "execute-task/M001/S01/T01",
+      "hook/review/M001/S01/T01",
+    ],
+    "stuck accounting should record the final replaced unit",
+  );
+  assert.ok(
+    !notifications.some((n) => n.includes("Worktree health check failed") || n.includes("Stuck on execute-task")),
+    "health and stuck notifications must use the final replaced unit",
+  );
+});
+
 test("autoLoop warns but proceeds for greenfield project (no project files) (#1833)", async () => {
   _resetPendingResolve();
 

package/src/resources/extensions/gsd/tests/auto-phases-lifecycle.test.ts

@@ -1,17 +1,20 @@
 import test from "node:test";
 import assert from "node:assert/strict";
+import { mkdtempSync, rmSync } from "node:fs";
+import { join } from "node:path";
+import { tmpdir } from "node:os";
 
 import { runFinalize } from "../auto/phases.ts";
 import { AutoSession } from "../auto/session.ts";
+import { readUnitRuntimeRecord, writeUnitRuntimeRecord } from "../unit-runtime.ts";
 
-test("runFinalize clears currentUnit after successful finalize", async () => {
-  const s = new AutoSession();
-  s.basePath = "/tmp/gsd-finalize-current-unit";
-  s.currentUnit = {
-    type: "execute-task",
-    id: "M001/S01/T01",
-    startedAt: Date.now(),
-  };
+async function runSuccessfulFinalize(s: AutoSession) {
+  const unit = s.currentUnit;
+  assert.ok(unit, "test setup must provide currentUnit");
+
+  writeUnitRuntimeRecord(s.basePath, unit.type, unit.id, unit.startedAt, {
+    phase: "dispatched",
+  });
 
   const deps = {
     clearUnitTimeout() {},
@@ -26,7 +29,7 @@ test("runFinalize clears currentUnit after successful finalize", async () => {
     postUnitPostVerification: async () => "continue",
   };
 
-  const result = await runFinalize(
+  return runFinalize(
     {
       ctx: { ui: { notify() {} } },
       pi: {},
@@ -38,8 +41,8 @@ test("runFinalize clears currentUnit after successful finalize", async () => {
       nextSeq: () => 1,
     } as any,
     {
-      unitType: "execute-task",
-      unitId: "M001/S01/T01",
+      unitType: unit.type,
+      unitId: unit.id,
       prompt: "",
       finalPrompt: "",
       pauseAfterUatDispatch: false,
@@ -55,7 +58,47 @@ test("runFinalize clears currentUnit after successful finalize", async () => {
       consecutiveFinalizeTimeouts: 0,
     },
   );
+}
+
+test("runFinalize clears currentUnit after successful finalize", async () => {
+  const base = mkdtempSync(join(tmpdir(), "gsd-finalize-current-unit-"));
+  const s = new AutoSession();
+  s.basePath = base;
+  s.currentUnit = {
+    type: "execute-task",
+    id: "M001/S01/T01",
+    startedAt: Date.now(),
+  };
+
+  try {
+    const result = await runSuccessfulFinalize(s);
+
+    assert.equal(result.action, "next");
+    assert.equal(s.currentUnit, null);
+  } finally {
+    rmSync(base, { recursive: true, force: true });
+  }
+});
+
+test("runFinalize marks unit runtime finalized after successful finalize", async () => {
+  const base = mkdtempSync(join(tmpdir(), "gsd-finalize-runtime-"));
+  const s = new AutoSession();
+  const startedAt = Date.now();
+  s.basePath = base;
+  s.currentUnit = {
+    type: "complete-milestone",
+    id: "M001",
+    startedAt,
+  };
+
+  try {
+    const result = await runSuccessfulFinalize(s);
+    const runtime = readUnitRuntimeRecord(base, "complete-milestone", "M001");
 
-  assert.equal(result.action, "next");
-  assert.equal(s.currentUnit, null);
+    assert.equal(result.action, "next");
+    assert.equal(runtime?.phase, "finalized");
+    assert.equal(runtime?.lastProgressKind, "finalize-success");
+  } finally {
+    rmSync(base, { recursive: true, force: true });
+  }
 });

package/src/resources/extensions/gsd/tests/auto-recovery.test.ts

@@ -5,7 +5,7 @@ import { join } from "node:path";
 import { tmpdir } from "node:os";
 import { randomUUID } from "node:crypto";
 
-import { verifyExpectedArtifact, hasImplementationArtifacts, resolveExpectedArtifactPath, diagnoseExpectedArtifact, buildLoopRemediationSteps, writeBlockerPlaceholder } from "../auto-recovery.ts";
+import { verifyExpectedArtifact, hasImplementationArtifacts, resolveExpectedArtifactPath, diagnoseExpectedArtifact, buildLoopRemediationSteps, writeBlockerPlaceholder, refreshRecoveryDbForArtifact } from "../auto-recovery.ts";
 import { resolveMilestoneFile } from "../paths.ts";
 import { openDatabase, closeDatabase, insertMilestone, insertSlice, insertGateRow, insertTask, getMilestoneCommitAttributionShas } from "../gsd-db.ts";
 import { clearParseCache } from "../files.ts";
@@ -175,6 +175,19 @@ test("resolveExpectedArtifactPath returns correct path for all slice-level types
   }
 });
 
+test("refreshRecoveryDbForArtifact treats missing execute-task DB rows as fatal mismatches", () => {
+  makeTmpProject();
+
+  const result = refreshRecoveryDbForArtifact("execute-task", "M001/S01/T01");
+
+  assert.deepEqual(result, {
+    ok: false,
+    fatal: true,
+    reason: "execute-task-artifact-db-missing",
+    message: "Stuck recovery found execute-task M001/S01/T01 artifacts, but no matching DB task row exists after refresh.",
+  });
+});
+
 // ─── diagnoseExpectedArtifact ─────────────────────────────────────────────
 
 test("diagnoseExpectedArtifact returns description for known types", () => {