gsd-pi 2.75.0-next.9c0ec0ba1 → 2.76.0-dev.b072ebb73

package/src/resources/extensions/gsd/setup-catalog.ts

@@ -0,0 +1,105 @@
+// GSD — Setup catalog (single source of truth for onboarding steps + provider sub-views)
+//
+// Re-exports filtered views over PROVIDER_REGISTRY (key-manager.ts) and owns the
+// canonical ONBOARDING_STEPS list. Consumers (CLI wizard, /gsd setup hub,
+// onboarding handler, web alignment) all read from here so adding a step or
+// provider lands in one place. Keep this module thin: no behavior beyond
+// filters + lookup helpers, so it stays cycle-safe even though it depends on
+// key-manager for the provider catalog.
+
+import { PROVIDER_REGISTRY, type ProviderInfo } from "./key-manager.js"
+
+export type OnboardingStepId =
+  | "llm"
+  | "model"
+  | "search"
+  | "remote"
+  | "tool-keys"
+  | "prefs"
+  | "skills"
+  | "doctor"
+  | "project"
+
+export interface OnboardingStepDef {
+  id: OnboardingStepId
+  label: string
+  /** Required steps gate the "complete" flag. Skipped required steps mark the wizard incomplete. */
+  required: boolean
+  /** Short description shown in /gsd setup status hub. */
+  hint: string
+}
+
+/**
+ * Canonical ordered list of onboarding steps.
+ *
+ * To add a new step:
+ *   1. Append here (or insert at the right position).
+ *   2. Bump FLOW_VERSION in onboarding-state.ts so existing users get re-prompted.
+ *   3. Wire its CLI runner in src/onboarding.ts (and handlers/onboarding.ts for --step).
+ */
+export const ONBOARDING_STEPS: readonly OnboardingStepDef[] = [
+  { id: "llm",       label: "LLM provider & auth",      required: true,  hint: "Sign in or paste an API key" },
+  { id: "model",     label: "Default model",            required: false, hint: "Pick a default model for the chosen provider" },
+  { id: "search",    label: "Web search provider",      required: false, hint: "Brave, Tavily, or Anthropic built-in" },
+  { id: "remote",    label: "Remote questions",         required: false, hint: "Discord / Slack / Telegram notifications" },
+  { id: "tool-keys", label: "Tool API keys",            required: false, hint: "Context7, Jina, Groq voice, etc." },
+  { id: "prefs",     label: "Global preferences",       required: false, hint: "Mode, profile, notifications" },
+  { id: "skills",    label: "Skills install",           required: false, hint: "Browse and install skill plugins" },
+  { id: "doctor",    label: "Validate setup",           required: false, hint: "Run provider doctor checks" },
+  { id: "project",   label: "Project init",             required: false, hint: "Bootstrap .gsd/ in this repo" },
+]
+
+const STEP_INDEX = new Map(ONBOARDING_STEPS.map((s, i) => [s.id, i]))
+
+export function getStep(id: string): OnboardingStepDef | undefined {
+  const idx = STEP_INDEX.get(id as OnboardingStepId)
+  return idx === undefined ? undefined : ONBOARDING_STEPS[idx]
+}
+
+export function isValidStepId(id: string): id is OnboardingStepId {
+  return STEP_INDEX.has(id as OnboardingStepId)
+}
+
+/**
+ * Given a possibly-stale resume point, return the nearest next step that is
+ * still defined in the catalog. Falls back to the first step.
+ */
+export function nearestResumeStep(lastResumePoint: string | null, completedSteps: string[]): OnboardingStepId {
+  const completed = new Set(completedSteps)
+  // First incomplete step at or after the lastResumePoint
+  let startIdx = 0
+  if (lastResumePoint && STEP_INDEX.has(lastResumePoint as OnboardingStepId)) {
+    startIdx = STEP_INDEX.get(lastResumePoint as OnboardingStepId) ?? 0
+  }
+  for (let i = startIdx; i < ONBOARDING_STEPS.length; i++) {
+    if (!completed.has(ONBOARDING_STEPS[i].id)) return ONBOARDING_STEPS[i].id
+  }
+  // Everything from the resume point is complete — try from the start
+  for (const step of ONBOARDING_STEPS) {
+    if (!completed.has(step.id)) return step.id
+  }
+  return ONBOARDING_STEPS[0].id
+}
+
+// ─── Provider catalog views ───────────────────────────────────────────────────
+
+export function getLlmProviders(): ProviderInfo[] {
+  return PROVIDER_REGISTRY.filter(p => p.category === "llm")
+}
+
+export function getToolProviders(): ProviderInfo[] {
+  return PROVIDER_REGISTRY.filter(p => p.category === "tool")
+}
+
+export function getSearchProviders(): ProviderInfo[] {
+  return PROVIDER_REGISTRY.filter(p => p.category === "search")
+}
+
+export function getRemoteProviders(): ProviderInfo[] {
+  return PROVIDER_REGISTRY.filter(p => p.category === "remote")
+}
+
+/** Provider IDs that count as "the user has an LLM configured" for shouldRunOnboarding. */
+export function getLlmProviderIds(): string[] {
+  return Array.from(new Set([...getLlmProviders().map(p => p.id), "claude-code"]))
+}

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,
@@ -610,7 +613,7 @@ async function handleAllSlicesDone(
   const validationTerminal = validationContent ? isValidationTerminal(validationContent) : false;
   const verdict = validationContent ? extractVerdict(validationContent) : undefined;
 
-  if (!validationTerminal || verdict === 'needs-remediation') {
+  if (!validationTerminal) {
     return {
       activeMilestone, activeSlice: null, activeTask: null,
       phase: 'validating-milestone',
@@ -621,6 +624,24 @@ async function handleAllSlicesDone(
     };
   }
 
+  // All roadmap slices are done (enforced by caller) and verdict is
+  // needs-remediation — remediation cannot progress without new slices.
+  // Return blocked instead of re-dispatching validate-milestone (#4506).
+  if (verdict === 'needs-remediation') {
+    return {
+      activeMilestone, activeSlice: null, activeTask: null,
+      phase: 'blocked',
+      recentDecisions: [],
+      blockers: [
+        `Milestone ${activeMilestone.id} validation verdict is needs-remediation but all slices are complete. ` +
+          `Add remediation slices via gsd_reassess_roadmap or override the verdict manually.`,
+      ],
+      nextAction: `Resolve ${activeMilestone.id} remediation before proceeding.`,
+      registry, requirements,
+      progress: { milestones: milestoneProgress, slices: sliceProgress },
+    };
+  }
+
   return {
     activeMilestone, activeSlice: null, activeTask: null,
     phase: 'completing-milestone',
@@ -849,7 +870,15 @@ 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 +898,24 @@ 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 +1008,29 @@ 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) {
@@ -1402,9 +1468,11 @@ export async function _deriveStateImpl(basePath: string): Promise<GSDState> {
       total: activeRoadmap.slices.length,
     };
 
-    // Force re-validation when verdict is needs-remediation — remediation slices
-    // may have completed since the stale validation was written (#3596).
-    if (!validationTerminal || verdict === 'needs-remediation') {
+    // Force re-validation when VALIDATION.md is absent or non-terminal —
+    // remediation slices may have completed since the stale validation was
+    // written (#3596). But needs-remediation with all slices done is a dead
+    // end — return blocked to avoid an infinite dispatch loop (#4506).
+    if (!validationTerminal) {
       return {
         activeMilestone,
         activeSlice: null,
@@ -1422,6 +1490,27 @@ export async function _deriveStateImpl(basePath: string): Promise<GSDState> {
       };
     }
 
+    if (verdict === 'needs-remediation') {
+      return {
+        activeMilestone,
+        activeSlice: null,
+        activeTask: null,
+        phase: 'blocked',
+        recentDecisions: [],
+        blockers: [
+          `Milestone ${activeMilestone.id} validation verdict is needs-remediation but all slices are complete. ` +
+            `Add remediation slices via gsd_reassess_roadmap or override the verdict manually.`,
+        ],
+        nextAction: `Resolve ${activeMilestone.id} remediation before proceeding.`,
+        registry,
+        requirements,
+        progress: {
+          milestones: milestoneProgress,
+          slices: sliceProgress,
+        },
+      };
+    }
+
     return {
       activeMilestone,
       activeSlice: null,

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

@@ -51,7 +51,7 @@ uok:
     enabled: true
   gitops:
     enabled: true
-    turn_action: status-only
+    turn_action: commit
     turn_push: false
   audit_unified:
     enabled: true

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-dashboard.test.ts

@@ -3,6 +3,7 @@ import assert from "node:assert/strict";
 import { mkdirSync, readFileSync, rmSync, writeFileSync } from "node:fs";
 import { join } from "node:path";
 import { tmpdir } from "node:os";
+import { execFileSync } from "node:child_process";
 
 import {
   unitVerb,
@@ -15,6 +16,10 @@ import {
   getWidgetMode,
   cycleWidgetMode,
   _resetWidgetModeForTests,
+  _resetLastCommitCacheForTests,
+  _refreshLastCommitForTests,
+  _getLastCommitForTests,
+  _getLastCommitFetchedAtForTests,
 } from "../auto-dashboard.ts";
 
 const autoSource = readFileSync(join(process.cwd(), "src", "resources", "extensions", "gsd", "auto.ts"), "utf-8");
@@ -215,6 +220,50 @@ test("auto progress widget renders RTK savings under the footer stats line", ()
   assert.match(dashboardSource, /lines\.push\(rightAlign\("", theme\.fg\("dim", cachedRtkLabel\), width\)\);/);
 });
 
+test("last commit refresh backs off cleanly when base path is not a git repo", (t) => {
+  const dir = makeTempDir("non-git");
+  mkdirSync(dir, { recursive: true });
+
+  t.after(() => {
+    cleanup(dir);
+    _resetLastCommitCacheForTests();
+  });
+
+  _resetLastCommitCacheForTests();
+  _refreshLastCommitForTests(dir);
+
+  assert.equal(_getLastCommitForTests(dir), null);
+  assert.ok(
+    _getLastCommitFetchedAtForTests() > 0,
+    "non-git refresh should still advance fetchedAt to avoid render-loop retries",
+  );
+});
+
+test("last commit refresh still returns commit info for a valid git repo", (t) => {
+  const dir = makeTempDir("git");
+  mkdirSync(dir, { recursive: true });
+
+  execFileSync("git", ["init", "-b", "main"], { cwd: dir, stdio: "pipe" });
+  execFileSync("git", ["config", "user.name", "GSD Test"], { cwd: dir, stdio: "pipe" });
+  execFileSync("git", ["config", "user.email", "gsd@example.com"], { cwd: dir, stdio: "pipe" });
+  writeFileSync(join(dir, "README.md"), "hello\n", "utf-8");
+  execFileSync("git", ["add", "README.md"], { cwd: dir, stdio: "pipe" });
+  execFileSync("git", ["commit", "-m", "test: seed dashboard repo"], { cwd: dir, stdio: "pipe" });
+
+  t.after(() => {
+    cleanup(dir);
+    _resetLastCommitCacheForTests();
+  });
+
+  _resetLastCommitCacheForTests();
+  _refreshLastCommitForTests(dir);
+
+  const lastCommit = _getLastCommitForTests(dir);
+  assert.ok(lastCommit, "git repo should produce last commit metadata");
+  assert.match(lastCommit!.message, /test: seed dashboard repo/);
+  assert.ok(lastCommit!.timeAgo.length > 0, "relative time should be populated");
+});
+
 // ─── extractUatSliceId ───────────────────────────────────────────────────
 
 test("extractUatSliceId extracts slice ID from M001/S01 format", () => {

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

@@ -0,0 +1,63 @@
+import test from "node:test";
+import assert from "node:assert/strict";
+import { mkdtempSync, mkdirSync, writeFileSync, existsSync, rmSync } from "node:fs";
+import { join } from "node:path";
+import { tmpdir } from "node:os";
+
+import { recoverFailedMigration } from "../migrate-external.ts";
+
+// Regression tests for #4416: `.gsd.migrating` must be healed before auto-mode
+// proceeds, including on the resume path in auto.ts (fixed at auto.ts:1325).
+// The `recoverFailedMigration` function is already called in `auto-start.ts:350`
+// for fresh-start sessions. The fix adds an identical call to `auto.ts:startAuto`
+// so that resume sessions (triggered by a persisted paused-session.json) also heal
+// a leftover `.gsd.migrating` directory before acquiring the session lock.
+
+test("recoverFailedMigration renames .gsd.migrating to .gsd when .gsd is absent", (t) => {
+  const base = mkdtempSync(join(tmpdir(), "gsd-migrating-recovery-"));
+  t.after(() => rmSync(base, { recursive: true, force: true }));
+
+  const migratingPath = join(base, ".gsd.migrating");
+  mkdirSync(join(migratingPath, "milestones"), { recursive: true });
+
+  const recovered = recoverFailedMigration(base);
+
+  assert.equal(recovered, true, "expected recovery to succeed");
+  assert.ok(existsSync(join(base, ".gsd")), ".gsd must exist after recovery");
+  assert.ok(!existsSync(migratingPath), ".gsd.migrating must not exist after recovery");
+});
+
+test("recoverFailedMigration returns false when .gsd.migrating is absent (nothing to do)", (t) => {
+  const base = mkdtempSync(join(tmpdir(), "gsd-migrating-noop-"));
+  t.after(() => rmSync(base, { recursive: true, force: true }));
+
+  const recovered = recoverFailedMigration(base);
+  assert.equal(recovered, false, "no migration to recover");
+});
+
+test("recoverFailedMigration returns false when both .gsd and .gsd.migrating exist (ambiguous)", (t) => {
+  const base = mkdtempSync(join(tmpdir(), "gsd-migrating-ambig-"));
+  t.after(() => rmSync(base, { recursive: true, force: true }));
+
+  mkdirSync(join(base, ".gsd"), { recursive: true });
+  mkdirSync(join(base, ".gsd.migrating"), { recursive: true });
+
+  const recovered = recoverFailedMigration(base);
+  assert.equal(recovered, false, "should not touch ambiguous state");
+  assert.ok(existsSync(join(base, ".gsd")), ".gsd must still exist");
+  assert.ok(existsSync(join(base, ".gsd.migrating")), ".gsd.migrating must still exist");
+});
+
+test("recoverFailedMigration preserves contents of .gsd.migrating", (t) => {
+  const base = mkdtempSync(join(tmpdir(), "gsd-migrating-contents-"));
+  t.after(() => rmSync(base, { recursive: true, force: true }));
+
+  const migratingPath = join(base, ".gsd.migrating");
+  mkdirSync(join(migratingPath, "milestones", "M001"), { recursive: true });
+  writeFileSync(join(migratingPath, "milestones", "M001", "M001-ROADMAP.md"), "# M001\n");
+
+  recoverFailedMigration(base);
+
+  const roadmap = join(base, ".gsd", "milestones", "M001", "M001-ROADMAP.md");
+  assert.ok(existsSync(roadmap), "Milestone file must be accessible via .gsd after recovery");
+});

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

@@ -7,7 +7,7 @@ import { fileURLToPath } from "node:url";
 
 const __dirname = dirname(fileURLToPath(import.meta.url));
 
-import { resolvePreferredModelConfig, resolveModelId } from "../auto-model-selection.js";
+import { resolvePreferredModelConfig, resolveModelId, selectAndApplyModel } from "../auto-model-selection.js";
 
 function makeTempDir(prefix: string): string {
   return mkdtempSync(join(tmpdir(), prefix));
@@ -46,6 +46,7 @@ test("resolvePreferredModelConfig synthesizes heavy routing ceiling when models
     assert.deepEqual(config, {
       primary: "claude-opus-4-6",
       fallbacks: [],
+      source: "synthesized",
     });
   } finally {
     process.chdir(originalCwd);
@@ -88,6 +89,7 @@ test("resolvePreferredModelConfig falls back to auto start model when heavy tier
     assert.deepEqual(config, {
       primary: "openai/gpt-5.4",
       fallbacks: [],
+      source: "synthesized",
     });
   } finally {
     process.chdir(originalCwd);
@@ -131,6 +133,7 @@ test("resolvePreferredModelConfig keeps explicit phase models as the ceiling", (
     assert.deepEqual(config, {
       primary: "claude-sonnet-4-6",
       fallbacks: [],
+      source: "explicit",
     });
   } finally {
     process.chdir(originalCwd);
@@ -141,6 +144,85 @@ test("resolvePreferredModelConfig keeps explicit phase models as the ceiling", (
   }
 });
 
+test("selectAndApplyModel honors explicit phase models without downgrading (#3617)", async () => {
+  const originalCwd = process.cwd();
+  const originalGsdHome = process.env.GSD_HOME;
+  const tempProject = makeTempDir("gsd-routing-project-");
+  const tempGsdHome = makeTempDir("gsd-routing-home-");
+  const setModelCalls: string[] = [];
+  let beforeModelSelectCalled = false;
+
+  try {
+    mkdirSync(join(tempProject, ".gsd"), { recursive: true });
+    writeFileSync(
+      join(tempProject, ".gsd", "PREFERENCES.md"),
+      [
+        "---",
+        "models:",
+        "  planning: claude-opus-4-6",
+        "dynamic_routing:",
+        "  enabled: true",
+        "  tier_models:",
+        "    light: gpt-4o-mini",
+        "    standard: claude-sonnet-4-6",
+        "    heavy: claude-opus-4-6",
+        "---",
+      ].join("\n"),
+      "utf-8",
+    );
+    process.env.GSD_HOME = tempGsdHome;
+    process.chdir(tempProject);
+
+    const availableModels = [
+      { id: "claude-opus-4-6", provider: "anthropic", api: "anthropic-messages" },
+      { id: "claude-sonnet-4-6", provider: "anthropic", api: "anthropic-messages" },
+      { id: "gpt-4o-mini", provider: "openai", api: "responses" },
+    ];
+
+    const result = await selectAndApplyModel(
+      {
+        modelRegistry: { getAvailable: () => availableModels },
+        sessionManager: { getSessionId: () => "test-session" },
+        ui: { notify: () => {} },
+        model: { provider: "anthropic", id: "claude-opus-4-6", api: "anthropic-messages" },
+      } as any,
+      {
+        setModel: async (model: { provider: string; id: string }) => {
+          setModelCalls.push(`${model.provider}/${model.id}`);
+          return true;
+        },
+        emitBeforeModelSelect: async () => {
+          beforeModelSelectCalled = true;
+          return undefined;
+        },
+        getActiveTools: () => [],
+        emitAdjustToolSet: async () => undefined,
+        setActiveTools: () => {},
+      } as any,
+      "plan-slice",
+      "slice-1",
+      tempProject,
+      undefined,
+      false,
+      { provider: "anthropic", id: "claude-opus-4-6" },
+      undefined,
+      true,
+    );
+
+    assert.equal(beforeModelSelectCalled, false, "explicit phase models should skip dynamic routing hooks");
+    assert.deepEqual(setModelCalls, ["anthropic/claude-opus-4-6"]);
+    assert.equal(result.routing, null, "explicit phase models should not record a routing downgrade");
+    assert.equal(result.appliedModel?.provider, "anthropic");
+    assert.equal(result.appliedModel?.id, "claude-opus-4-6");
+  } finally {
+    process.chdir(originalCwd);
+    if (originalGsdHome === undefined) delete process.env.GSD_HOME;
+    else process.env.GSD_HOME = originalGsdHome;
+    rmSync(tempProject, { recursive: true, force: true });
+    rmSync(tempGsdHome, { recursive: true, force: true });
+  }
+});
+
 // ─── resolveModelId tests ─────────────────────────────────────────────────
 
 test("resolveModelId: bare ID resolves to claude-code when session is claude-code (#3772)", () => {
@@ -247,6 +329,18 @@ test("model policy receives task metadata for requirement-vector decisions", ()
   );
 });
 
+test("dynamic routing passes provider-qualified model keys to the router", () => {
+  const src = readFileSync(join(__dirname, "..", "auto-model-selection.ts"), "utf-8");
+  assert.ok(
+    src.includes("routingEligibleModels.map(m => `${m.provider}/${m.id}`)"),
+    "selectAndApplyModel should preserve provider prefixes for dynamic routing candidates",
+  );
+  assert.ok(
+    !src.includes("routingEligibleModels.map(m => m.id)"),
+    "selectAndApplyModel must not strip providers before resolving tier_models",
+  );
+});
+
 test("resolveModelId: anthropic wins over claude-code when session provider is not claude-code", () => {
   const availableModels = [
     { id: "claude-sonnet-4-6", provider: "claude-code" },

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

@@ -0,0 +1,35 @@
+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 { buildSourceFilePaths } from "../auto-prompts.ts";
+
+// Regression test for #4416: the fallback string must not mention `rg` because
+// auto-mode runs on systems where ripgrep is not installed (e.g. Windows).
+test("buildSourceFilePaths fallback does not reference rg or ripgrep", (t) => {
+  const tmp = mkdtempSync(join(tmpdir(), "gsd-prompts-fallback-"));
+  t.after(() => rmSync(tmp, { recursive: true, force: true }));
+
+  // No GSD files exist in tmp — forces the fallback branch.
+  const result = buildSourceFilePaths(tmp, "M001");
+
+  assert.ok(
+    !result.includes("rg ") && !result.includes("`rg`") && !result.includes("ripgrep"),
+    `Fallback string must not reference rg/ripgrep. Got: ${result}`,
+  );
+  assert.ok(result.length > 0, "Fallback string must not be empty");
+});
+
+test("buildSourceFilePaths with sid also produces rg-free fallback", (t) => {
+  const tmp = mkdtempSync(join(tmpdir(), "gsd-prompts-fallback-sid-"));
+  t.after(() => rmSync(tmp, { recursive: true, force: true }));
+
+  const result = buildSourceFilePaths(tmp, "M001", "S01");
+
+  assert.ok(
+    !result.includes("rg ") && !result.includes("`rg`") && !result.includes("ripgrep"),
+    `Fallback string must not reference rg/ripgrep. Got: ${result}`,
+  );
+});