gsd-pi 2.82.0 → 3.0.0

package/src/resources/extensions/claude-code-cli/tests/partial-builder.test.ts

@@ -1,7 +1,24 @@
 import { describe, test } from "node:test";
 import assert from "node:assert/strict";
-import { mapContentBlock, parseMcpToolName, PartialMessageBuilder } from "../partial-builder.ts";
-import type { BetaContentBlock, BetaRawMessageStreamEvent } from "../sdk-types.ts";
+import { mapContentBlock, mapUsage, parseMcpToolName, PartialMessageBuilder } from "../partial-builder.ts";
+import type { BetaContentBlock, BetaRawMessageStreamEvent, NonNullableUsage } from "../sdk-types.ts";
+
+describe("mapUsage", () => {
+	test("excludes cumulative cache reads from context-sized totalTokens (#5243)", () => {
+		const usage: NonNullableUsage = {
+			input_tokens: 150_000,
+			output_tokens: 2_000,
+			cache_read_input_tokens: 900_000,
+			cache_creation_input_tokens: 3_000,
+		};
+
+		const mapped = mapUsage(usage, 1.23);
+
+		assert.equal(mapped.cacheRead, 900_000);
+		assert.equal(mapped.totalTokens, 155_000);
+		assert.equal(mapped.cost.total, 1.23);
+	});
+});
 
 describe("PartialMessageBuilder — malformed tool arguments (#2574)", () => {
 	/**

package/src/resources/extensions/claude-code-cli/tests/stream-adapter.test.ts

@@ -1258,9 +1258,18 @@ describe("stream-adapter — abort classification (F2)", () => {
 	test("recognizes Claude Code SDK abort exceptions", () => {
 		assert.equal(isClaudeCodeAbortErrorMessage("Claude Code process aborted by user"), true);
 		assert.equal(isClaudeCodeAbortErrorMessage("Request aborted by user"), true);
+		assert.equal(isClaudeCodeAbortErrorMessage("AbortError: The operation was aborted"), true);
 		assert.equal(isClaudeCodeAbortErrorMessage("rate limit exceeded"), false);
 	});
 
+	test("does not misclassify non-user abort contexts", () => {
+		assert.equal(isClaudeCodeAbortErrorMessage("Job aborted due to timeout"), false);
+		assert.equal(isClaudeCodeAbortErrorMessage("Operation aborted: disk full"), false);
+		assert.equal(isClaudeCodeAbortErrorMessage("aborted by system cleanup"), false);
+		assert.equal(isClaudeCodeAbortErrorMessage("Database transaction aborted due to constraint violation"), false);
+		assert.equal(isClaudeCodeAbortErrorMessage("Connection aborted unexpectedly"), false);
+	});
+
 	test("makeAbortedMessage sets stopReason to 'aborted', not 'error'", () => {
 		const message = makeAbortedMessage("claude-sonnet-4-6", "");
 		assert.equal(message.stopReason, "aborted");

package/src/resources/extensions/cmux/index.ts

@@ -374,6 +374,12 @@ export class CmuxClient {
     const stdout = await this.runAsync(["send-surface", "--surface", surfaceId, payload]);
     return stdout !== null;
   }
+
+  // Send Ctrl-C (ETX) to a surface to interrupt the running command.
+  async sendInterrupt(surfaceId: string): Promise<boolean> {
+    const stdout = await this.runAsync(["send-surface", "--surface", surfaceId, "\x03"]);
+    return stdout !== null;
+  }
 }
 
 export function syncCmuxSidebar(preferences: CmuxPreferences | undefined, state: CmuxState): void {

package/src/resources/extensions/gsd/auto/contracts.ts

@@ -2,27 +2,31 @@
 // File Purpose: Auto Orchestration module interfaces and ADR-015 invariant adapter contracts.
 
 import type { GSDState } from "../types.js";
+import type { MinimalModelRegistry } from "../context-budget.js";
 
 export interface AutoSessionContext {
   basePath: string;
   trigger: "guided-flow" | "resume" | "auto-loop" | "manual";
 }
 
+export interface UnitRef {
+  unitType: string;
+  unitId: string;
+}
+
 export interface AutoStatus {
   phase: "idle" | "running" | "paused" | "stopped" | "error";
-  activeUnit?: {
-    unitType: string;
-    unitId: string;
-  };
+  activeUnit?: UnitRef;
   lastTransitionAt?: number;
   transitionCount: number;
 }
 
-export interface AutoAdvanceResult {
-  kind: "advanced" | "blocked" | "paused" | "stopped" | "error";
-  reason?: string;
-  stateSnapshot?: GSDState;
-}
+export type AutoAdvanceResult =
+  | { kind: "advanced"; unit: UnitRef; stateSnapshot: GSDState }
+  | { kind: "blocked"; reason: string; action: "pause" | "stop"; stateSnapshot?: GSDState }
+  | { kind: "stopped"; reason: string; stateSnapshot?: GSDState }
+  | { kind: "paused"; reason: string }
+  | { kind: "error"; reason: string };
 
 export interface AutoOrchestrationModule {
   start(sessionContext: AutoSessionContext): Promise<AutoAdvanceResult>;
@@ -33,12 +37,30 @@ export interface AutoOrchestrationModule {
 }
 
 export interface DispatchAdapter {
-  decideNextUnit(input: { stateSnapshot: GSDState }): Promise<{
-    unitType: string;
-    unitId: string;
-    reason: string;
-    preconditions: string[];
-  } | null>;
+  decideNextUnit(input: {
+    stateSnapshot: GSDState;
+    /** Mirrors `DispatchContext.structuredQuestionsAvailable` — "true"/"false" string per the dispatch contract. */
+    structuredQuestionsAvailable?: "true" | "false";
+    /** Session model context window in tokens, forwarded to the budget engine. */
+    sessionContextWindow?: number;
+    /** Session model provider, used for provider-specific effective context windows. */
+    sessionProvider?: string;
+    /** Model registry for executor-model lookups inside the budget engine. */
+    modelRegistry?: MinimalModelRegistry;
+  }): Promise<
+    | {
+        kind: "blocked";
+        reason: string;
+        action: "pause" | "stop";
+      }
+    | {
+        unitType: string;
+        unitId: string;
+        reason: string;
+        preconditions: string[];
+      }
+    | null
+  >;
 }
 
 export interface RecoveryAdapter {
@@ -70,11 +92,31 @@ export interface WorktreeAdapter {
   cleanupOnStop(reason: string): Promise<void>;
 }
 
+export type HealthGateResult =
+  | { kind: "pass"; fixesApplied?: readonly string[] }
+  | { kind: "fail"; reason: string }
+  | { kind: "threw"; error: unknown };
+
 export interface HealthAdapter {
-  preAdvanceGate(): Promise<{ allow: boolean; reason?: string }>;
+  checkResourcesStale(): string | null;
+  preAdvanceGate(): Promise<HealthGateResult>;
   postAdvanceRecord(result: AutoAdvanceResult): Promise<void>;
 }
 
+export interface UokGateInput {
+  gateId: string;
+  gateType: "policy" | "execution";
+  outcome: "pass" | "fail" | "manual-attention";
+  failureClass: "none" | "policy" | "manual-attention";
+  rationale: string;
+  findings?: string;
+  milestoneId?: string;
+}
+
+export interface UokGateAdapter {
+  emit(input: UokGateInput): Promise<void>;
+}
+
 export interface RuntimePersistenceAdapter {
   ensureLockOwnership(): Promise<void>;
   journalTransition(event: {
@@ -101,4 +143,5 @@ export interface AutoOrchestratorDeps {
   health: HealthAdapter;
   runtime: RuntimePersistenceAdapter;
   notifications: NotificationAdapter;
+  uokGate: UokGateAdapter;
 }

package/src/resources/extensions/gsd/auto/infra-errors.ts

@@ -7,9 +7,14 @@
  */
 
 /**
- * Error codes indicating infrastructure failures that cannot be recovered by
- * retrying. Each retry re-dispatches the unit at full LLM cost, so we bail
- * immediately rather than burning budget on guaranteed failures.
+ * Error codes indicating infrastructure-level failures from the OS,
+ * filesystem, or network. This set includes permanent resource failures
+ * (ENOSPC, ENOMEM, EROFS), transient resource exhaustion (EAGAIN, ENOBUFS),
+ * and network/offline errors (ECONNREFUSED, ENOTFOUND, ENETUNREACH).
+ *
+ * Transient git failures are retried separately through
+ * TRANSIENT_GIT_RETRY_CODES in native-git-bridge.ts before escalating to the
+ * auto-loop.
  */
 export const INFRA_ERROR_CODES: ReadonlySet<string> = new Set([
   "ENOSPC",   // disk full
@@ -19,6 +24,7 @@ export const INFRA_ERROR_CODES: ReadonlySet<string> = new Set([
   "EMFILE",   // too many open files (process)
   "ENFILE",   // too many open files (system)
   "EAGAIN",       // resource temporarily unavailable (resource exhaustion)
+  "ENOBUFS",      // no buffer space available (transient pipe exhaustion)
   "ECONNREFUSED", // connection refused (offline / local server down)
   "ENOTFOUND",    // DNS lookup failed (offline / no network)
   "ENETUNREACH",  // network unreachable (offline / no route)

package/src/resources/extensions/gsd/auto/loop.ts

@@ -1,3 +1,5 @@
+// Project/App: GSD-2
+// File Purpose: Main auto-mode execution loop.
 /**
  * auto/loop.ts — Main auto-mode execution loop.
  *
@@ -78,7 +80,10 @@ import { createWorkflowTurnReporter } from "./workflow-turn-reporter.js";
 import { validateWorkflowSessionLock } from "./workflow-session-lock.js";
 import { dequeueSidecarItem } from "./workflow-sidecar-queue.js";
 import { maintainWorkerHeartbeat } from "./workflow-worker-heartbeat.js";
-import { measureMemoryPressure } from "./workflow-memory-pressure.js";
+import {
+  measureMemoryPressure,
+  shouldCheckMemoryPressure,
+} from "./workflow-memory-pressure.js";
 import { buildSidecarIterationData } from "./workflow-sidecar-iteration.js";
 import {
   createExecutionGraphUnitDispatchDeps,
@@ -203,9 +208,9 @@ function logCustomVerifyRetrySaveFailure(err: unknown): void {
 }
 
 // ── Memory pressure monitoring (#3331) ──────────────────────────────────
-// Check heap usage every N iterations and trigger graceful shutdown before
-// the OS OOM killer sends SIGKILL. The threshold is 90% of the V8 heap
-// limit (--max-old-space-size or default ~1.5-4GB depending on platform).
+// Check heap usage on session startup, then every N iterations, and trigger
+// graceful shutdown before the OS OOM killer sends SIGKILL. The threshold is
+// 90% of the V8 heap limit (--max-old-space-size or default ~1.5-4GB depending on platform).
 const MEMORY_CHECK_INTERVAL = 5; // check every 5 iterations
 const MAX_CUSTOM_ENGINE_VERIFY_RETRIES = 3;
 
@@ -372,7 +377,7 @@ export async function autoLoop(
 
     // ── Memory pressure check (#3331) ──
     // Graceful shutdown before OOM killer sends SIGKILL.
-    if (iteration % MEMORY_CHECK_INTERVAL === 0) {
+    if (shouldCheckMemoryPressure(iteration, MEMORY_CHECK_INTERVAL)) {
       const mem = measureMemoryPressure();
       debugLog("autoLoop", { phase: "memory-check", ...mem });
       const memoryDecision = decideMemoryPressure({ ...mem, iteration });
@@ -942,11 +947,18 @@ export async function autoLoop(
         unitId: iterData.unitId,
       });
       const finalizeReason = finalizeResult.action === "break" ? finalizeResult.reason : undefined;
+      const finalizeStatus = finalizeReason === "step-wizard"
+        ? "completed"
+        : finalizeResult.action === "next"
+          ? "completed"
+          : finalizeResult.action === "continue"
+            ? "retry"
+            : "stopped";
       journalReporter.emit("post-unit-finalize-end", {
         iteration,
         unitType: iterData.unitType,
         unitId: iterData.unitId,
-        status: finalizeResult.action === "next" ? "completed" : finalizeResult.action === "continue" ? "retry" : "stopped",
+        status: finalizeStatus,
         action: finalizeResult.action,
         ...(finalizeReason ? { reason: finalizeReason } : {}),
       });
@@ -993,6 +1005,10 @@ export async function autoLoop(
       }) || dispatchSettled;
       completeIteration();
       finishTurn("completed");
+      if (finalizeDecision.action === "complete-and-break") {
+        s.preserveStepSurfaceAfterLoopExit = true;
+        break;
+      }
     } catch (loopErr) {
       // ── Blanket catch: absorb unexpected exceptions, apply graduated recovery ──
       const msg = loopErr instanceof Error ? loopErr.message : String(loopErr);

package/src/resources/extensions/gsd/auto/orchestrator.ts

@@ -7,6 +7,16 @@ function now(): number {
   return Date.now();
 }
 
+/**
+ * Size of the dispatch-decision ring buffer used by the Auto Orchestration
+ * module's stuck-loop detector. When the same `${unitType}:${unitId}` key
+ * fills the window, advance() blocks with `action: "stop"`.
+ *
+ * Mirrors the legacy `STUCK_WINDOW_SIZE` in auto/phases.ts so behaviour is
+ * preserved across the eventual cutover (issue #5791).
+ */
+export const STUCK_WINDOW_SIZE = 6;
+
 export class AutoOrchestrator implements AutoOrchestrationModule {
   private status: AutoStatus = {
     phase: "idle",
@@ -14,6 +24,7 @@ export class AutoOrchestrator implements AutoOrchestrationModule {
   };
   private readonly deps: AutoOrchestratorDeps;
   private lastAdvanceKey: string | null = null;
+  private dispatchKeyWindow: string[] = [];
 
   public constructor(deps: AutoOrchestratorDeps) {
     this.deps = deps;
@@ -21,6 +32,7 @@ export class AutoOrchestrator implements AutoOrchestrationModule {
 
   public async start(_sessionContext: AutoSessionContext): Promise<AutoAdvanceResult> {
     this.lastAdvanceKey = null;
+    this.dispatchKeyWindow = [];
     this.status.phase = "running";
     this.bumpTransition();
     await this.deps.runtime.journalTransition({ name: "start" });
@@ -31,19 +43,72 @@ export class AutoOrchestrator implements AutoOrchestrationModule {
   public async advance(): Promise<AutoAdvanceResult> {
     try {
       await this.deps.runtime.ensureLockOwnership();
+
+      const staleMsg = this.deps.health.checkResourcesStale();
+      if (staleMsg) {
+        await this.deps.uokGate.emit({
+          gateId: "resource-version-guard",
+          gateType: "policy",
+          outcome: "fail",
+          failureClass: "policy",
+          rationale: "resource version guard blocked dispatch",
+          findings: staleMsg,
+        });
+        const blocked: AutoAdvanceResult = { kind: "blocked", reason: staleMsg, action: "stop" };
+        await this.deps.runtime.journalTransition({ name: "advance-blocked", reason: blocked.reason });
+        await this.deps.health.postAdvanceRecord(blocked);
+        return blocked;
+      }
+      await this.deps.uokGate.emit({
+        gateId: "resource-version-guard",
+        gateType: "policy",
+        outcome: "pass",
+        failureClass: "none",
+        rationale: "resource version guard passed",
+      });
+
       const gate = await this.deps.health.preAdvanceGate();
-      if (!gate.allow) {
-        const blocked: AutoAdvanceResult = { kind: "blocked", reason: gate.reason ?? "health gate blocked" };
+      if (gate.kind === "fail") {
+        await this.deps.uokGate.emit({
+          gateId: "pre-dispatch-health-gate",
+          gateType: "execution",
+          outcome: "manual-attention",
+          failureClass: "manual-attention",
+          rationale: "pre-dispatch health gate blocked dispatch",
+          findings: gate.reason,
+        });
+        const blocked: AutoAdvanceResult = { kind: "blocked", reason: gate.reason, action: "pause" };
         await this.deps.runtime.journalTransition({ name: "advance-blocked", reason: blocked.reason });
         await this.deps.health.postAdvanceRecord(blocked);
         return blocked;
       }
+      if (gate.kind === "threw") {
+        await this.deps.uokGate.emit({
+          gateId: "pre-dispatch-health-gate",
+          gateType: "execution",
+          outcome: "manual-attention",
+          failureClass: "manual-attention",
+          rationale: "pre-dispatch health gate threw unexpectedly",
+          findings: String(gate.error),
+        });
+        // intentional fall-through: matches runPreDispatch behaviour
+      } else {
+        await this.deps.uokGate.emit({
+          gateId: "pre-dispatch-health-gate",
+          gateType: "execution",
+          outcome: "pass",
+          failureClass: "none",
+          rationale: "pre-dispatch health gate passed",
+          findings: gate.fixesApplied?.join(", ") ?? "",
+        });
+      }
 
       const reconciliation = await this.deps.stateReconciliation.reconcileBeforeDispatch();
       if (!reconciliation.ok || !reconciliation.stateSnapshot) {
         const blocked: AutoAdvanceResult = {
           kind: "blocked",
-          reason: reconciliation.reason,
+          reason: reconciliation.reason ?? "state reconciliation produced no snapshot",
+          action: "pause",
           stateSnapshot: reconciliation.stateSnapshot,
         };
         await this.deps.runtime.journalTransition({ name: "advance-blocked", reason: blocked.reason });
@@ -57,15 +122,64 @@ export class AutoOrchestrator implements AutoOrchestrationModule {
         this.status.phase = "stopped";
         this.status.activeUnit = undefined;
         this.lastAdvanceKey = null;
+        this.dispatchKeyWindow = [];
         this.bumpTransition();
         await this.deps.runtime.journalTransition({ name: "advance-stopped", reason: stopped.reason });
         await this.deps.health.postAdvanceRecord(stopped);
         return stopped;
       }
+      if (!("unitType" in decision)) {
+        const blocked: AutoAdvanceResult = {
+          kind: "blocked",
+          reason: decision.reason,
+          action: decision.action,
+          stateSnapshot: reconciliation.stateSnapshot,
+        };
+        await this.deps.runtime.journalTransition({ name: "advance-blocked", reason: blocked.reason });
+        await this.deps.health.postAdvanceRecord(blocked);
+        return blocked;
+      }
 
       const nextKey = `${decision.unitType}:${decision.unitId}`;
-      if (this.lastAdvanceKey === nextKey) {
-        const blocked: AutoAdvanceResult = { kind: "blocked", reason: "idempotent advance: unit already active" };
+
+      // Record every dispatch decision in the ring buffer before pre-flight
+      // checks so the stuck-loop detector observes the full decision history
+      // (including decisions that idempotency would otherwise short-circuit).
+      // The ring is capped at STUCK_WINDOW_SIZE and evicts oldest-first.
+      this.dispatchKeyWindow.push(nextKey);
+      if (this.dispatchKeyWindow.length > STUCK_WINDOW_SIZE) {
+        this.dispatchKeyWindow.shift();
+      }
+
+      // Idempotency: same key as immediately previous successful advance.
+      // This is the soft, fast-path block kept from #5786. It only fires when
+      // the ring is NOT yet saturated for this key — once the ring is full of
+      // `nextKey`, the stuck-loop verdict takes precedence (see below). Both
+      // checks coexist: idempotency for the common immediate-repeat case,
+      // stuck-loop for the saturated-window case.
+      const matchingCount = this.dispatchKeyWindow.filter((k) => k === nextKey).length;
+      if (this.lastAdvanceKey === nextKey && matchingCount < STUCK_WINDOW_SIZE) {
+        const blocked: AutoAdvanceResult = { kind: "blocked", reason: "idempotent advance: unit already active", action: "stop" };
+        await this.deps.runtime.journalTransition({
+          name: "advance-blocked",
+          reason: blocked.reason,
+          unitType: decision.unitType,
+          unitId: decision.unitId,
+        });
+        await this.deps.health.postAdvanceRecord(blocked);
+        return blocked;
+      }
+
+      // Stuck-loop detection: when the ring is saturated with copies of
+      // `nextKey` (count >= STUCK_WINDOW_SIZE), the orchestrator has been
+      // picking the same unit across the whole window and must hard-stop with
+      // a diagnosable reason.
+      if (matchingCount >= STUCK_WINDOW_SIZE) {
+        const blocked: AutoAdvanceResult = {
+          kind: "blocked",
+          reason: `stuck-loop: ${nextKey} picked ${matchingCount} times`,
+          action: "stop",
+        };
         await this.deps.runtime.journalTransition({
           name: "advance-blocked",
           reason: blocked.reason,
@@ -81,6 +195,7 @@ export class AutoOrchestrator implements AutoOrchestrationModule {
         const blocked: AutoAdvanceResult = {
           kind: "blocked",
           reason: contract.reason,
+          action: "pause",
           stateSnapshot: reconciliation.stateSnapshot,
         };
         await this.deps.runtime.journalTransition({
@@ -98,6 +213,7 @@ export class AutoOrchestrator implements AutoOrchestrationModule {
         const blocked: AutoAdvanceResult = {
           kind: "blocked",
           reason: worktree.reason,
+          action: "pause",
           stateSnapshot: reconciliation.stateSnapshot,
         };
         await this.deps.runtime.journalTransition({
@@ -123,7 +239,11 @@ export class AutoOrchestrator implements AutoOrchestrationModule {
       });
       await this.deps.worktree.syncAfterUnit(decision.unitType, decision.unitId);
 
-      const advanced: AutoAdvanceResult = { kind: "advanced", stateSnapshot: reconciliation.stateSnapshot };
+      const advanced: AutoAdvanceResult = {
+        kind: "advanced",
+        unit: { unitType: decision.unitType, unitId: decision.unitId },
+        stateSnapshot: reconciliation.stateSnapshot,
+      };
       await this.deps.health.postAdvanceRecord(advanced);
       return advanced;
     } catch (error) {
@@ -148,6 +268,7 @@ export class AutoOrchestrator implements AutoOrchestrationModule {
 
       if (result.kind === "stopped") {
         this.lastAdvanceKey = null;
+        this.dispatchKeyWindow = [];
         this.status.activeUnit = undefined;
       }
       this.bumpTransition();
@@ -173,6 +294,7 @@ export class AutoOrchestrator implements AutoOrchestrationModule {
 
   public async resume(): Promise<AutoAdvanceResult> {
     this.lastAdvanceKey = null;
+    this.dispatchKeyWindow = [];
     this.status.phase = "running";
     this.bumpTransition();
     await this.deps.runtime.journalTransition({ name: "resume" });
@@ -188,6 +310,7 @@ export class AutoOrchestrator implements AutoOrchestrationModule {
     this.status.phase = "stopped";
     this.status.activeUnit = undefined;
     this.lastAdvanceKey = null;
+    this.dispatchKeyWindow = [];
     this.bumpTransition();
     await this.deps.runtime.journalTransition({ name: "stop", reason });
     await this.deps.notifications.notifyLifecycle({ name: "stop", detail: reason });