gsd-pi 2.75.0-dev.063e5a3 → 2.75.0-dev.fd2382b9f

package/src/resources/extensions/gsd/run-manager.ts

@@ -16,8 +16,9 @@
 import { mkdirSync, writeFileSync, existsSync, readdirSync, statSync } from "node:fs";
 import { join } from "node:path";
 import { stringify } from "yaml";
-import { loadDefinition, substituteParams } from "./definition-loader.js";
+import { loadDefinition, loadDefinitionFromFile, substituteParams } from "./definition-loader.js";
 import { initializeGraph, writeGraph, readGraph } from "./graph.js";
+import { resolvePlugin } from "./workflow-plugins.js";
 import type { WorkflowDefinition } from "./definition-loader.js";
 import type { WorkflowGraph } from "./graph.js";
 
@@ -67,48 +68,81 @@ function deriveStatus(graph: WorkflowGraph): "pending" | "running" | "complete"
 
 // ─── Public API ──────────────────────────────────────────────────────────
 
+/**
+ * Create a run directory from an explicit definition file path.
+ * Preferred over `createRun` when the caller has already resolved the file
+ * (e.g. via the plugin resolver).
+ */
+export function createRunFromDefinition(
+  basePath: string,
+  defName: string,
+  definitionFile: string,
+  overrides?: Record<string, string>,
+): string {
+  const rawDef = loadDefinitionFromFile(definitionFile);
+  const def: WorkflowDefinition = overrides
+    ? substituteParams(rawDef, overrides)
+    : substituteParams(rawDef);
+
+  const timestamp = makeTimestamp();
+  const runDir = join(basePath, ".gsd", RUNS_DIR, defName, timestamp);
+  mkdirSync(runDir, { recursive: true });
+
+  writeFileSync(join(runDir, "DEFINITION.yaml"), stringify(def), "utf-8");
+
+  const graph = initializeGraph(def);
+  writeGraph(runDir, graph);
+
+  if (overrides && Object.keys(overrides).length > 0) {
+    writeFileSync(
+      join(runDir, "PARAMS.json"),
+      JSON.stringify(overrides, null, 2),
+      "utf-8",
+    );
+  }
+
+  return runDir;
+}
+
 /**
  * Create a new isolated run directory for a workflow definition.
  *
- * 1. Loads the definition from `<basePath>/.gsd/workflow-defs/<defName>.yaml`
- * 2. Applies parameter substitution if overrides are provided
- * 3. Creates `<basePath>/.gsd/workflow-runs/<defName>/<timestamp>/`
- * 4. Writes frozen DEFINITION.yaml, initialized GRAPH.yaml, and optional PARAMS.json
+ * Resolution order:
+ *   1. Plugin resolver (project → global → bundled), YAML format only.
+ *   2. Legacy `.gsd/workflow-defs/<defName>.yaml`.
  *
- * @param basePath — project root directory
- * @param defName — definition filename (without .yaml extension)
- * @param overrides — optional parameter overrides (merged with definition defaults)
- * @returns Full path to the created run directory
- * @throws Error if the definition file doesn't exist or is invalid
+ * Creates `<basePath>/.gsd/workflow-runs/<defName>/<timestamp>/` containing
+ * DEFINITION.yaml (frozen), GRAPH.yaml (initialized), and optional PARAMS.json.
+ *
+ * @throws Error if no matching definition is found anywhere.
  */
 export function createRun(
   basePath: string,
   defName: string,
   overrides?: Record<string, string>,
 ): string {
-  const defsDir = join(basePath, ".gsd", DEFS_DIR);
+  // Try the unified plugin resolver first — honors project/global overrides.
+  const plugin = resolvePlugin(basePath, defName);
+  if (plugin && plugin.format === "yaml") {
+    return createRunFromDefinition(basePath, defName, plugin.path, overrides);
+  }
 
-  // Load and validate the definition
+  // Fall back to legacy `.gsd/workflow-defs/<defName>.yaml`.
+  const defsDir = join(basePath, ".gsd", DEFS_DIR);
   const rawDef = loadDefinition(defsDir, defName);
-
-  // Apply parameter substitution if overrides provided
   const def: WorkflowDefinition = overrides
     ? substituteParams(rawDef, overrides)
-    : substituteParams(rawDef); // still resolve default params if any
+    : substituteParams(rawDef);
 
-  // Create the run directory
   const timestamp = makeTimestamp();
   const runDir = join(basePath, ".gsd", RUNS_DIR, defName, timestamp);
   mkdirSync(runDir, { recursive: true });
 
-  // Freeze the definition as DEFINITION.yaml
   writeFileSync(join(runDir, "DEFINITION.yaml"), stringify(def), "utf-8");
 
-  // Initialize and write GRAPH.yaml
   const graph = initializeGraph(def);
   writeGraph(runDir, graph);
 
-  // Write PARAMS.json if overrides were provided
   if (overrides && Object.keys(overrides).length > 0) {
     writeFileSync(
       join(runDir, "PARAMS.json"),

package/src/resources/extensions/gsd/state.ts

@@ -44,6 +44,8 @@ import { existsSync, readdirSync, readFileSync } from 'node:fs';
 import { debugCount, debugTime } from './debug-logger.js';
 import { logWarning, logError } from './workflow-logger.js';
 import { extractVerdict } from './verdict-parser.js';
+import { loadEffectiveGSDPreferences } from './preferences.js';
+import { detectPendingEscalation } from './escalation.js';
 
 import {
   isDbAvailable,
@@ -60,6 +62,7 @@ import {
   updateSliceStatus,
   updateTaskStatus,
   getPendingGateCountForTurn,
+  autoHealSketchFlags,
   type MilestoneRow,
   type SliceRow,
   type TaskRow,
@@ -849,7 +852,15 @@ export async function deriveStateFromDb(basePath: string): Promise<GSDState> {
     return handleAllSlicesDone(basePath, activeMilestone, registry, requirements, milestoneProgress, sliceProgress);
   }
 
-  const activeSliceContext = resolveSliceDependencies(activeMilestoneSlices);
+  // ADR-011 auto-heal: if a slice has a PLAN on disk but is still flagged is_sketch=1
+  // (e.g. a crash between plan-slice write and the sketch flip), reconcile before
+  // running phase derivation so the flag doesn't misroute state.
+  autoHealSketchFlags(activeMilestone.id, (sid) =>
+    !!resolveSliceFile(basePath, activeMilestone.id, sid, "PLAN"),
+  );
+  // Re-read slices after auto-heal so downstream reads see fresh is_sketch values.
+  const healedSlices = getMilestoneSlices(activeMilestone.id);
+  const activeSliceContext = resolveSliceDependencies(healedSlices);
   if (!activeSliceContext.activeSlice) {
     // If locked slice wasn't found, it returns null but logs warning, we need to return 'blocked'
     if (process.env.GSD_SLICE_LOCK) {
@@ -869,10 +880,24 @@ export async function deriveStateFromDb(basePath: string): Promise<GSDState> {
       progress: { milestones: milestoneProgress, slices: sliceProgress },
     };
   }
-  const { activeSlice } = activeSliceContext;
+  const { activeSlice, activeSliceRow } = activeSliceContext;
 
   const planFile = resolveSliceFile(basePath, activeMilestone.id, activeSlice.id, "PLAN");
   if (!planFile) {
+    // ADR-011: sketch slices with progressive_planning enabled enter the
+    // `refining` phase — a refine-slice unit expands the sketch into a full plan
+    // before execution. When the flag is off, sketches are indistinguishable
+    // from a missing plan and fall through to the normal `planning` phase.
+    const progressive = loadEffectiveGSDPreferences()?.preferences?.phases?.progressive_planning === true;
+    if (progressive && activeSliceRow?.is_sketch === 1) {
+      return {
+        activeMilestone, activeSlice, activeTask: null,
+        phase: 'refining', recentDecisions: [], blockers: [],
+        nextAction: `Refine sketch slice ${activeSlice.id} (${activeSlice.title}) using prior slice context.`,
+        registry, requirements,
+        progress: { milestones: milestoneProgress, slices: sliceProgress },
+      };
+    }
     return {
       activeMilestone, activeSlice, activeTask: null,
       phase: 'planning', recentDecisions: [], blockers: [],
@@ -965,6 +990,29 @@ export async function deriveStateFromDb(basePath: string): Promise<GSDState> {
     }
   }
 
+  // ADR-011 Phase 2: pause-on-escalation takes precedence over dispatching the
+  // next task. `awaiting_review` tasks (continueWithDefault=true) are NOT
+  // surfaced here — they let the loop continue.
+  //
+  // We do NOT gate this on `phases.mid_execution_escalation` — creation of
+  // new escalations is gated at the write site (tools/complete-task.ts:315),
+  // but any escalation_pending row already persisted in the DB must be
+  // honored even if the user later toggles the flag off. Otherwise those
+  // rows would silently orphan, the loop would advance past the paused task,
+  // and the user's prior resolution never lands.
+  const escalatingTaskId = detectPendingEscalation(tasks, basePath);
+  if (escalatingTaskId) {
+    return {
+      activeMilestone, activeSlice, activeTask,
+      phase: 'escalating-task', recentDecisions: [],
+      blockers: [`Task ${escalatingTaskId} requires a user decision before the loop can proceed`],
+      nextAction: `Run /gsd escalate show ${escalatingTaskId} to review, then /gsd escalate resolve ${escalatingTaskId} <choice> to proceed.`,
+      activeWorkspace: undefined,
+      registry, requirements,
+      progress: { milestones: milestoneProgress, slices: sliceProgress, tasks: taskProgress },
+    };
+  }
+
   if (!blockerTaskId) {
     const isTriggered = checkReplanTrigger(basePath, activeMilestone.id, activeSlice.id);
     if (isTriggered) {

package/src/resources/extensions/gsd/templates/PREFERENCES.md

@@ -44,19 +44,19 @@ uok:
   legacy_fallback:
     enabled: false
   gates:
-    enabled: false
+    enabled: true
   model_policy:
-    enabled: false
+    enabled: true
   execution_graph:
-    enabled: false
+    enabled: true
   gitops:
-    enabled: false
-    turn_action: status-only
+    enabled: true
+    turn_action: commit
     turn_push: false
   audit_unified:
-    enabled: false
+    enabled: true
   plan_v2:
-    enabled: false
+    enabled: true
 auto_visualize:
 auto_report:
 parallel:

package/src/resources/extensions/gsd/tests/agent-end-retry.test.ts

@@ -2,8 +2,7 @@
  * agent-end-retry.test.ts — Regression checks for the agent_end model.
  *
  * The per-unit one-shot resolve function lives at module level in auto-loop.ts
- * (_currentResolve). handleAgentEnd is a thin compatibility wrapper around
- * resolveAgentEnd().
+ * (_currentResolve). agent_end is handled via resolveAgentEnd().
  */
 
 import test from "node:test";
@@ -62,38 +61,6 @@ test("legacy pendingAgentEndRetry state is gone", () => {
   );
 });
 
-test("handleAgentEnd is a thin compatibility wrapper", () => {
-  const source = getAutoTsSource();
-  const fnIdx = source.indexOf("export async function handleAgentEnd");
-  assert.ok(fnIdx > -1, "handleAgentEnd must exist in auto.ts");
-  const fnBlock = source.slice(fnIdx, source.indexOf("\n// ─── ", fnIdx + 100));
-
-  assert.ok(
-    fnBlock.includes("resolveAgentEnd("),
-    "handleAgentEnd must delegate to resolveAgentEnd",
-  );
-  assert.ok(
-    !fnBlock.includes("pendingAgentEndRetry"),
-    "handleAgentEnd must not use legacy retry state",
-  );
-  assert.ok(
-    !fnBlock.includes("dispatchNextUnit"),
-    "handleAgentEnd must not dispatch recursively",
-  );
-});
-
-test("handleAgentEnd early return calls resolveAgentEndCancelled", () => {
-  const source = getAutoTsSource();
-  const fnIdx = source.indexOf("export async function handleAgentEnd");
-  assert.ok(fnIdx > -1, "handleAgentEnd must exist in auto.ts");
-  const fnBlock = source.slice(fnIdx, source.indexOf("\n// ─── ", fnIdx + 100));
-
-  assert.ok(
-    fnBlock.includes("resolveAgentEndCancelled()"),
-    "handleAgentEnd must call resolveAgentEndCancelled on early return to prevent orphaned promises",
-  );
-});
-
 test("pauseAuto calls resolveAgentEndCancelled to unblock the loop", () => {
   const source = getAutoTsSource();
   const fnIdx = source.indexOf("export async function pauseAuto");

package/src/resources/extensions/gsd/tests/artifact-corruption-2630.test.ts

@@ -46,6 +46,8 @@ function makeSliceRow(overrides?: Partial<SliceRow>): SliceRow {
     observability_impact: '',
     sequence: 4,
     replan_triggered_at: null,
+    is_sketch: 0,
+    sketch_scope: '',
     ...overrides,
   };
 }
@@ -77,6 +79,11 @@ function makeTaskRow(overrides?: Partial<TaskRow>): TaskRow {
     expected_output: [],
     observability_impact: '',
     sequence: 0,
+    blocker_source: '',
+    escalation_pending: 0,
+    escalation_awaiting_review: 0,
+    escalation_artifact_path: null,
+    escalation_override_applied_at: null,
     ...overrides,
   };
 }

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

@@ -313,6 +313,42 @@ test("runUnit re-applies the selected unit model after newSession before dispatc
   assert.equal(pi.calls.length, 1);
 });
 
+test("runUnit cancels before dispatch when model restore fails after newSession", async () => {
+  _resetPendingResolve();
+
+  const notifications: Array<{ message: string; level: string }> = [];
+  const ctx = makeMockCtx();
+  ctx.ui.notify = (message: string, level: string) => {
+    notifications.push({ message, level });
+  };
+
+  const pi = makeMockPi();
+  pi.setModel = async (...args: unknown[]) => {
+    pi.setModelCalls.push(args);
+    return false;
+  };
+
+  const s = makeMockSession();
+  s.currentUnitModel = { provider: "openai-codex", id: "gpt-5.4" };
+
+  const result = await runUnit(ctx, pi, s, "task", "T01", "prompt");
+
+  assert.equal(result.status, "cancelled");
+  assert.equal(result.errorContext?.category, "session-failed");
+  assert.match(
+    result.errorContext?.message ?? "",
+    /Failed to restore configured model openai-codex\/gpt-5\.4 after session creation/,
+  );
+  assert.equal(pi.setModelCalls.length, 1);
+  assert.equal(pi.calls.length, 0, "unit must not dispatch on the session default model");
+  assert.deepEqual(notifications, [
+    {
+      message: "Failed to restore configured model openai-codex/gpt-5.4 after session creation. Cancelling unit before dispatch.",
+      level: "warning",
+    },
+  ]);
+});
+
 // ─── Structural assertions ───────────────────────────────────────────────────
 
 test("auto-loop.ts exports autoLoop, runUnit, resolveAgentEnd", async () => {
@@ -1267,7 +1303,7 @@ test("auto-loop.ts barrel re-exports autoLoop, runUnit, and resolveAgentEnd", ()
   );
 });
 
-test("auto.ts startAuto dispatches through the UOK kernel wrapper (legacy loop adapter)", () => {
+test("auto.ts startAuto dispatches through the UOK kernel wrapper with explicit kernel and legacy paths", () => {
   const src = readFileSync(
     resolve(import.meta.dirname, "..", "auto.ts"),
     "utf-8",
@@ -1283,8 +1319,12 @@ test("auto.ts startAuto dispatches through the UOK kernel wrapper (legacy loop a
     "startAuto must dispatch through runAutoLoopWithUok()",
   );
   assert.ok(
-    fnBlock.includes("runLegacyLoop: autoLoop"),
-    "startAuto must preserve the legacy autoLoop adapter in kernel dispatch",
+    fnBlock.includes("runKernelLoop: runUokKernelLoop"),
+    "startAuto must wire the explicit UOK kernel loop path",
+  );
+  assert.ok(
+    fnBlock.includes("runLegacyLoop: runLegacyAutoLoop"),
+    "startAuto must preserve explicit legacy fallback dispatch",
   );
 });
 
@@ -1332,7 +1372,7 @@ test("startAuto guards against concurrent invocation (#2923)", () => {
   );
 });
 
-test("agent_end handler calls resolveAgentEnd (not handleAgentEnd)", () => {
+test("agent_end handler calls resolveAgentEnd (not the legacy auto.ts path)", () => {
   const hooksSrc = readFileSync(
     resolve(import.meta.dirname, "..", "bootstrap", "register-hooks.ts"),
     "utf-8",
@@ -1347,7 +1387,7 @@ test("agent_end handler calls resolveAgentEnd (not handleAgentEnd)", () => {
   );
   assert.ok(
     recoverySrc.includes("resolveAgentEnd(event)"),
-    "agent_end success path must call resolveAgentEnd(event) instead of handleAgentEnd(ctx, pi)",
+    "agent_end success path must call resolveAgentEnd(event) instead of legacy wrappers",
   );
   assert.ok(
     recoverySrc.includes("isSessionSwitchInFlight()"),
@@ -1389,32 +1429,6 @@ test("auto-timeout-recovery.ts calls resolveAgentEnd instead of dispatchNextUnit
   );
 });
 
-test("handleAgentEnd in auto.ts is a thin wrapper calling resolveAgentEnd", () => {
-  const src = readFileSync(
-    resolve(import.meta.dirname, "..", "auto.ts"),
-    "utf-8",
-  );
-  const fnIdx = src.indexOf("export async function handleAgentEnd");
-  assert.ok(fnIdx > -1, "handleAgentEnd must exist");
-  const fnEnd = src.indexOf("\n// ─── ", fnIdx + 100);
-  const fnBlock =
-    fnEnd > -1 ? src.slice(fnIdx, fnEnd) : src.slice(fnIdx, fnIdx + 1000);
-  assert.ok(
-    fnBlock.includes("resolveAgentEnd("),
-    "handleAgentEnd must call resolveAgentEnd",
-  );
-  // The function should be short — no reentrancy guard, no verification, no dispatch
-  assert.ok(
-    !fnBlock.includes("dispatchNextUnit"),
-    "handleAgentEnd must not call dispatchNextUnit (it's now a thin wrapper)",
-  );
-  assert.ok(
-    !fnBlock.includes("postUnitPreVerification") &&
-      !fnBlock.includes("postUnitPostVerification"),
-    "handleAgentEnd must not contain verification logic (moved to autoLoop)",
-  );
-});
-
 // ── Stuck counter tests ──────────────────────────────────────────────────────
 
 test("stuck detection: stops when sliding window detects same unit 3 consecutive times", async () => {

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

@@ -5,7 +5,8 @@ import { join } from "node:path";
 import { tmpdir } from "node:os";
 import { randomUUID } from "node:crypto";
 
-import { verifyExpectedArtifact, hasImplementationArtifacts, resolveExpectedArtifactPath, diagnoseExpectedArtifact, buildLoopRemediationSteps } from "../auto-recovery.ts";
+import { verifyExpectedArtifact, hasImplementationArtifacts, resolveExpectedArtifactPath, diagnoseExpectedArtifact, buildLoopRemediationSteps, writeBlockerPlaceholder } from "../auto-recovery.ts";
+import { resolveMilestoneFile } from "../paths.ts";
 import { openDatabase, closeDatabase, insertMilestone, insertSlice, insertGateRow } from "../gsd-db.ts";
 import { clearParseCache } from "../files.ts";
 import { parseRoadmap } from "../parsers-legacy.ts";
@@ -712,3 +713,124 @@ test("verifyExpectedArtifact checks pending gate-evaluate artifacts without ESM
 
   assert.equal(verified, false, "pending gates should keep gate-evaluate unverified");
 });
+
+// ─── #4414 regressions ────────────────────────────────────────────────────────
+
+test("#4414: writeBlockerPlaceholder invalidates path cache so dispatch guard sees file", () => {
+  const base = makeTmpBase();
+  try {
+    // Prime the readdir cache by resolving a DIFFERENT file first — this
+    // mirrors the stuck-loop condition where the dispatch guard cached an
+    // empty directory listing before the placeholder was written.
+    invalidateAllCaches();
+    assert.equal(
+      resolveMilestoneFile(base, "M001", "RESEARCH"),
+      null,
+      "no RESEARCH file yet",
+    );
+
+    const result = writeBlockerPlaceholder(
+      "research-milestone",
+      "M001",
+      base,
+      "verification retries exhausted",
+    );
+    assert.ok(result, "placeholder path returned");
+
+    // After writeBlockerPlaceholder, the dispatch guard must see the new file
+    // immediately — otherwise the rule re-fires (#4414, 7× re-dispatch).
+    const postResolve = resolveMilestoneFile(base, "M001", "RESEARCH");
+    assert.ok(
+      postResolve,
+      "resolveMilestoneFile finds the placeholder post-write (cache invalidated)",
+    );
+  } finally {
+    cleanup(base);
+  }
+});
+
+test("#4414: parallel-research sentinel path does not collide with RESEARCH suffix", () => {
+  const base = makeTmpBase();
+  try {
+    // Write only the parallel-research blocker (sentinel).
+    const sentinel = resolveExpectedArtifactPath(
+      "research-slice",
+      "M001/parallel-research",
+      base,
+    );
+    assert.ok(sentinel, "sentinel path resolves for parallel-research");
+    writeFileSync(sentinel!, "# blocker\n", "utf-8");
+
+    // Critical: the sentinel filename must NOT be matched by the legacy regex
+    // used when callers look up milestone-level RESEARCH. Otherwise the
+    // dispatch guard for research-milestone would short-circuit falsely.
+    const milestoneResearch = resolveMilestoneFile(base, "M001", "RESEARCH");
+    assert.equal(
+      milestoneResearch,
+      null,
+      "sentinel must not be mistaken for M001-RESEARCH.md via legacy pattern match",
+    );
+  } finally {
+    cleanup(base);
+  }
+});
+
+test("#4414: verifyExpectedArtifact parallel-research succeeds when all research-ready slices have RESEARCH", () => {
+  const base = makeTmpBase();
+  try {
+    mkdirSync(join(base, ".gsd", "milestones", "M001", "slices", "S02", "tasks"), { recursive: true });
+    mkdirSync(join(base, ".gsd", "milestones", "M001", "slices", "S03", "tasks"), { recursive: true });
+
+    // Minimal roadmap with three slices
+    writeFileSync(
+      join(base, ".gsd", "milestones", "M001", "M001-ROADMAP.md"),
+      [
+        "# M001: Regression",
+        "",
+        "## Slices",
+        "",
+        "- [ ] **S01: Alpha** `risk:low` `depends:[]`",
+        "- [ ] **S02: Beta** `risk:low` `depends:[]`",
+        "- [ ] **S03: Gamma** `risk:low` `depends:[]`",
+        "",
+      ].join("\n"),
+      "utf-8",
+    );
+
+    // Only 2 of 3 have RESEARCH — should fail verification
+    writeFileSync(
+      join(base, ".gsd", "milestones", "M001", "slices", "S01", "S01-RESEARCH.md"),
+      "# research",
+      "utf-8",
+    );
+    writeFileSync(
+      join(base, ".gsd", "milestones", "M001", "slices", "S02", "S02-RESEARCH.md"),
+      "# research",
+      "utf-8",
+    );
+
+    clearParseCache();
+    invalidateAllCaches();
+    assert.equal(
+      verifyExpectedArtifact("research-slice", "M001/parallel-research", base),
+      false,
+      "missing S03 RESEARCH → verification fails",
+    );
+
+    // All three RESEARCH present → verification passes
+    writeFileSync(
+      join(base, ".gsd", "milestones", "M001", "slices", "S03", "S03-RESEARCH.md"),
+      "# research",
+      "utf-8",
+    );
+    clearParseCache();
+    invalidateAllCaches();
+    assert.equal(
+      verifyExpectedArtifact("research-slice", "M001/parallel-research", base),
+      true,
+      "all slices have RESEARCH → verification passes",
+    );
+  } finally {
+    cleanup(base);
+  }
+});

package/src/resources/extensions/gsd/tests/auto-wrapup-inflight-guard.test.ts

@@ -15,6 +15,9 @@ const autoSrc = readFileSync(autoPath, "utf-8");
 const runUnitPath = join(import.meta.dirname, "..", "auto", "run-unit.ts");
 const runUnitSrc = readFileSync(runUnitPath, "utf-8");
 
+const registerHooksPath = join(import.meta.dirname, "..", "bootstrap", "register-hooks.ts");
+const registerHooksSrc = readFileSync(registerHooksPath, "utf-8");
+
 describe("#3512: gsd-auto-wrapup must not interrupt in-flight tool calls", () => {
   test("soft timeout wrapup gates triggerTurn on getInFlightToolCount() === 0", () => {
     // The soft timeout sendMessage must NOT use a hardcoded `triggerTurn: true`.
@@ -70,6 +73,42 @@ describe("#3512: gsd-auto-wrapup must not interrupt in-flight tool calls", () =>
   });
 });
 
+describe("#4276: pending/skipped tools stay visible to auto-mode hooks", () => {
+  test("tool_call handler marks GSD tools in-flight before execution_start", () => {
+    const startMarker = 'pi.on("tool_call", async (event, ctx) => {';
+    const endMarker = 'pi.on("tool_result", async (event) => {';
+    const toolCallSection = registerHooksSrc.slice(
+      registerHooksSrc.indexOf(startMarker),
+      registerHooksSrc.indexOf(endMarker),
+    );
+
+    assert.ok(toolCallSection.length > 0, "Could not locate tool_call handler section");
+    assert.ok(
+      toolCallSection.includes("markToolStart(event.toolCallId, event.toolName)"),
+      "tool_call handler must mark tools pending before tool_execution_start fires",
+    );
+  });
+
+  test("tool_result handler clears pending tools and records queued-skip errors", () => {
+    const startMarker = 'pi.on("tool_result", async (event) => {';
+    const endMarker = 'pi.on("tool_execution_start", async (event) => {';
+    const toolResultSection = registerHooksSrc.slice(
+      registerHooksSrc.indexOf(startMarker),
+      registerHooksSrc.indexOf(endMarker),
+    );
+
+    assert.ok(toolResultSection.length > 0, "Could not locate tool_result handler section");
+    assert.ok(
+      toolResultSection.includes("markToolEnd(event.toolCallId)"),
+      "tool_result handler must clear pending tool tracking even when execution hooks never fire",
+    );
+    assert.ok(
+      toolResultSection.includes("recordToolInvocationError(event.toolName, errorText)"),
+      "tool_result handler must surface queued-skip errors for GSD tools",
+    );
+  });
+});
+
 describe("#3512: pauseAuto and stopAuto must flush queued follow-up messages", () => {
   test("stopAuto calls clearQueue()", () => {
     // stopAuto must flush queued messages to prevent late async_job_result

package/src/resources/extensions/gsd/tests/autocomplete-regressions-1675.test.ts

@@ -44,6 +44,14 @@ test("/gsd description includes discuss", () => {
   );
 });
 
+test("/gsd description includes debug", () => {
+  const pi = createMockPi();
+  registerGSDCommand(pi as any);
+
+  const gsd = pi.commands.get("gsd");
+  assert.ok(gsd.description.includes("debug"), "description should include debug");
+});
+
 test("/gsd next completions include --debug", () => {
   const pi = createMockPi();
   registerGSDCommand(pi as any);
@@ -54,6 +62,18 @@ test("/gsd next completions include --debug", () => {
   assert.ok(debug, "next --debug should appear in completions");
 });
 
+test("/gsd debug completions include list|status|continue|--diagnose", () => {
+  const pi = createMockPi();
+  registerGSDCommand(pi as any);
+
+  const gsd = pi.commands.get("gsd");
+  const completions = gsd.getArgumentCompletions("debug ");
+  const values = completions.map((c: any) => c.value);
+  for (const expected of ["debug list", "debug status", "debug continue", "debug --diagnose"]) {
+    assert.ok(values.includes(expected), `missing completion: ${expected}`);
+  }
+});
+
 test("/gsd widget completions include full|small|min|off", () => {
   const pi = createMockPi();
   registerGSDCommand(pi as any);
@@ -66,6 +86,25 @@ test("/gsd widget completions include full|small|min|off", () => {
   }
 });
 
+test("/gsd logs completions still include debug after adding /gsd debug", () => {
+  const pi = createMockPi();
+  registerGSDCommand(pi as any);
+
+  const gsd = pi.commands.get("gsd");
+  const completions = gsd.getArgumentCompletions("logs ");
+  const values = completions.map((c: any) => c.value);
+  assert.ok(values.includes("logs debug"), "logs debug completion should remain available");
+});
+
+test("/gsd help full includes /gsd debug command", async () => {
+  const ctx = createMockCtx();
+
+  await handleGSDCommand("help full", ctx as any, {} as any);
+
+  const helpText = ctx.notifications.map((n) => n.message).join("\n");
+  assert.match(helpText, /\/gsd debug\s+Create\/list\/continue persistent debug sessions/);
+});
+
 test("bare /gsd skip shows usage and does not fall through to unknown-command warning", async () => {
   const ctx = createMockCtx();