@opengsd/gsd-pi 1.2.0-dev.4c756166 → 1.2.0-dev.6ccd27b3

package/src/resources/extensions/gsd/auto-post-unit.ts

@@ -18,7 +18,7 @@ import { deriveState } from "./state.js";
 import { logWarning, logError } from "./workflow-logger.js";
 import { loadFile, parseSummary, resolveAllOverrides } from "./files.js";
 import { loadPrompt } from "./prompt-loader.js";
-import { isAwaitingUserInput } from "./user-input-boundary.js";
+import { isAwaitingUserInput } from "./consent-question.js";
 import {
   resolveMilestonePath,
   resolveSliceFile,
@@ -51,8 +51,7 @@ import { createWorkspace, scopeMilestone } from "./workspace.js";
 import { normalizeWorktreePathForCompare } from "./worktree-root.js";
 import { isDbAvailable, getTask, getSlice, getMilestone, getMilestoneSlices, updateTaskStatus, _getAdapter, getVerificationEvidence } from "./gsd-db.js";
 import { getWorkflowDatabasePath, refreshWorkflowDatabaseFromDisk } from "./db-workspace.js";
-import { renderPlanCheckboxes, renderRoadmapFromDb } from "./markdown-renderer.js";
-import { parseRoadmap as parseLegacyRoadmap } from "./parsers-legacy.js";
+import { renderPlanCheckboxes, renderRoadmapFromDb, roadmapRenderMarksSliceDone } from "./markdown-renderer.js";
 import { consumeSignal } from "./session-status-io.js";
 import {
   checkPostUnitHooks,
@@ -72,7 +71,7 @@ import {
 } from "./milestone-closeout.js";
 import type { AutoSession, SidecarItem } from "./auto/session.js";
 import { getEvidence, clearEvidenceFromDisk, isExecutionToolName } from "./safety/evidence-collector.js";
-import { validateFileChanges } from "./safety/file-change-validator.js";
+import { validateFileChanges, effectiveFileChangeAllowlist } from "./safety/file-change-validator.js";
 import { crossReferenceEvidence, type ClaimedEvidence } from "./safety/evidence-cross-ref.js";
 import { validateContent } from "./safety/content-validator.js";
 import { resolveSafetyHarnessConfig } from "./safety/safety-harness.js";
@@ -88,7 +87,7 @@ import { writeTurnGitTransaction } from "./uok/gitops.js";
 import { isClosedStatus } from "./status-guards.js";
 import { detectAbandonMilestone } from "./abandon-detect.js";
 import { getPendingGate } from "./bootstrap/write-gate.js";
-import { isDeterministicPolicyError } from "./auto-tool-tracking.js";
+import { isDeterministicPolicyError, isToolUnavailableError } from "./auto-tool-tracking.js";
 import { formatConnectedStepStack, formatPostUnitStatusCard } from "./auto-status-message.js";
 import {
   clearProjectResearchInflightMarker,
@@ -96,6 +95,7 @@ import {
 } from "./project-research-policy.js";
 import { validateArtifact } from "./schemas/validate.js";
 import { verificationRetryKey } from "./auto/verification-retry-policy.js";
+import { saveCustomVerifyRetryCounts } from "./auto/custom-verify-retry-store.js";
 import { getLedger } from "./metrics.js";
 import { getUnitCostSpikeAction, resolveUnitCostSpikeMultiplier } from "./auto-budget.js";
 import { resolveCanonicalMilestoneRoot } from "./worktree-manager.js";
@@ -619,8 +619,10 @@ export function _hasExecutionToolCallsInSessionForTest(entries: readonly unknown
       return true;
     }
 
-    if (e?.type !== "message") continue;
-    const msg = e?.message;
+    // Accept both session-manager entries ({type: "message", message}) and
+    // bare agent-end messages ({role, content}) — the auto loop passes the
+    // latter via opts.agentEndMessages.
+    const msg = e?.type === "message" ? e?.message : e;
     if (!msg || msg.role !== "assistant" || !Array.isArray(msg.content)) continue;
     for (const block of msg.content) {
       if (block?.type !== "toolCall") continue;
@@ -899,7 +901,7 @@ export const USER_DRIVEN_DEEP_UNITS = new Set([
   "discuss-milestone",
   "research-decision",
 ]);
-export { isAwaitingUserInput } from "./user-input-boundary.js";
+export { isAwaitingUserInput } from "./consent-question.js";
 
 function artifactValidationKind(unitType: string): "project" | "requirements" | null {
   if (unitType === "discuss-project") return "project";
@@ -988,11 +990,13 @@ async function repairCompleteSliceRoadmapProjection(
     return false;
   }
 
+  // Stale-render detection (ADR-017): the DB already says the slice is closed;
+  // this only checks whether the rendered ROADMAP projection reflects it, to
+  // decide whether a repair re-render is needed.
   const roadmapPath = resolveMilestoneFile(artifactBase, mid, "ROADMAP");
   if (roadmapPath && existsSync(roadmapPath)) {
     try {
-      const roadmap = parseLegacyRoadmap(readFileSync(roadmapPath, "utf-8"));
-      if (roadmap.slices.find((roadmapSlice) => roadmapSlice.id === sid)?.done) {
+      if (roadmapRenderMarksSliceDone(readFileSync(roadmapPath, "utf-8"), sid)) {
         return false;
       }
     } catch (err) {
@@ -1535,6 +1539,11 @@ export async function postUnitPreVerification(pctx: PostUnitContext, opts?: PreV
       if (safetyConfig.enabled) {
         const { milestone: sMid, slice: sSid, task: sTid } = parseUnitId(s.currentUnit.id);
 
+        const fileChangeAllowlist = effectiveFileChangeAllowlist(
+          safetyConfig.file_change_allowlist,
+          (prefs?.git as { manage_gitignore?: boolean } | undefined)?.manage_gitignore,
+        );
+
         // File change validation (execute-task only, after unit execution)
         if (safetyConfig.file_change_validation && s.currentUnit.type === "execute-task" && sMid && sSid && sTid) {
           try {
@@ -1554,7 +1563,7 @@ export async function postUnitPreVerification(pctx: PostUnitContext, opts?: PreV
                   files: taskRow.files,
                 })),
               );
-              const audit = validateFileChanges(s.basePath, expectedOutput, plannedFiles, safetyConfig.file_change_allowlist);
+              const audit = validateFileChanges(s.basePath, expectedOutput, plannedFiles, fileChangeAllowlist);
               if (audit && audit.violations.length > 0) {
                 const warnings = audit.violations.filter(v => v.severity === "warning");
                 for (const v of warnings) {
@@ -1576,7 +1585,7 @@ export async function postUnitPreVerification(pctx: PostUnitContext, opts?: PreV
                   s.basePath,
                   expectedOutput,
                   plannedFiles,
-                  safetyConfig.file_change_allowlist,
+                  fileChangeAllowlist,
                 );
                 if (audit && audit.violations.length > 0) {
                   const warnings = audit.violations.filter(v => v.severity === "warning");
@@ -2015,7 +2024,32 @@ export async function postUnitPreVerification(pctx: PostUnitContext, opts?: PreV
           "error",
         );
       } else if (!triggerArtifactVerified) {
-        if (s.lastToolInvocationError) {
+        if (s.lastToolInvocationError && isToolUnavailableError(s.lastToolInvocationError)) {
+          // Tool-unavailable is transient: the workflow MCP server registers
+          // its surface asynchronously, so a Unit's first call can race the
+          // registration. Retry with escalating delay, bounded at 3 attempts.
+          // ponytail: MAX constant so the guard, log, and display all agree
+          const MAX_TOOL_UNAVAIL_RETRIES = 3;
+          if (s.toolUnavailableRetries >= MAX_TOOL_UNAVAIL_RETRIES) {
+            debugLog("postUnit", { phase: "tool-unavailable-exhausted", unitType: s.currentUnit.type, unitId: s.currentUnit.id, retries: s.toolUnavailableRetries });
+            ctx.ui.notify(
+              `Tool unavailable for ${s.currentUnit.type} after ${MAX_TOOL_UNAVAIL_RETRIES} retries: ${s.lastToolInvocationError}. MCP server may not be starting — pausing auto-mode.`,
+              "error",
+            );
+            s.lastToolInvocationError = null;
+            await pauseAuto(ctx, pi);
+            return "dispatched";
+          }
+          s.toolUnavailableRetries++;
+          const delayMs = s.toolUnavailableRetries * 1000;
+          debugLog("postUnit", { phase: "tool-unavailable-retry", unitType: s.currentUnit.type, unitId: s.currentUnit.id, error: s.lastToolInvocationError, attempt: s.toolUnavailableRetries, delayMs });
+          ctx.ui.notify(
+            `Tool unavailable for ${s.currentUnit.type}: ${s.lastToolInvocationError}. Waiting ${delayMs}ms for MCP server — retry ${s.toolUnavailableRetries}/${MAX_TOOL_UNAVAIL_RETRIES}.`,
+            "warning",
+          );
+          s.lastToolInvocationError = null;
+          await new Promise(r => setTimeout(r, delayMs));
+        } else if (s.lastToolInvocationError) {
           const isUserSkip = /queued user message/i.test(s.lastToolInvocationError);
           const errMsg = isUserSkip
             ? `Tool skipped for ${s.currentUnit.type}: ${s.lastToolInvocationError}. Queued user message interrupted the turn — pausing auto-mode.`
@@ -2141,6 +2175,7 @@ export async function postUnitPreVerification(pctx: PostUnitContext, opts?: PreV
               }
             }
             s.exhaustedVerificationUnits.add(retryKey);
+            saveCustomVerifyRetryCounts(s, { logFailure: err => debugLog("postUnit", { phase: "save-verify-retries-failed", error: err instanceof Error ? err.message : String(err) }) });
             debugLog("postUnit", { phase: "artifact-verify-exhausted", unitType: s.currentUnit.type, unitId: s.currentUnit.id, attempt });
             ctx.ui.notify(
               `${failureDetails} Pausing auto-mode after ${MAX_ARTIFACT_VERIFICATION_RETRIES} retries.`,
@@ -2150,6 +2185,7 @@ export async function postUnitPreVerification(pctx: PostUnitContext, opts?: PreV
             return "dispatched";
           }
           s.verificationRetryCount.set(retryKey, attempt);
+          saveCustomVerifyRetryCounts(s, { logFailure: err => debugLog("postUnit", { phase: "save-verify-retries-failed", error: err instanceof Error ? err.message : String(err) }) });
           s.pendingVerificationRetry = {
             unitId: s.currentUnit.id,
             failureContext: `${failureDetails} (attempt ${attempt}/${MAX_ARTIFACT_VERIFICATION_RETRIES}).`,
@@ -2171,8 +2207,11 @@ export async function postUnitPreVerification(pctx: PostUnitContext, opts?: PreV
         if (s.pendingVerificationRetry?.unitId === s.currentUnit.id) {
           s.pendingVerificationRetry = null;
         }
+        s.toolUnavailableRetries = 0;
         s.verificationRetryCount.delete(retryKey);
         s.verificationRetryFailureHashes.delete(retryKey);
+        s.exhaustedVerificationUnits.delete(retryKey);
+        saveCustomVerifyRetryCounts(s, { logFailure: err => debugLog("postUnit", { phase: "save-verify-retries-failed", error: err instanceof Error ? err.message : String(err) }) });
 
         if (s.currentUnit.type === "complete-milestone") {
           const { milestone: mid } = parseUnitId(s.currentUnit.id);

package/src/resources/extensions/gsd/auto-prompts.ts

@@ -40,6 +40,7 @@ import {
   composeContextModeInstructions,
   composeContractedUnitContext,
   composeInlinedContext,
+  composeToolSurfaceInstructions,
   composeUnitContext,
   type ArtifactResolver,
   type ComposedUnitContextBlock,
@@ -57,7 +58,11 @@ import { debugLog } from "./debug-logger.js";
 import { buildSkillActivationBlock, buildSkillDiscoveryVars } from "./skill-activation.js";
 import { findMilestoneIds } from "./milestone-ids.js";
 import { buildRunUatPresentationForType, RUN_UAT_TOOL_PRESENTATION_PLAN_ID } from "./tool-presentation-plan.js";
-import { resolveEffectiveUatType, shouldDispatchUatForContent, type UatType } from "./uat-policy.js";
+import {
+  classifyUatContentForRun,
+  shouldDispatchUatForContent,
+  type UatType,
+} from "./uat-policy.js";
 import { buildWebAppUatGuidanceBlock } from "./web-app-uat.js";
 
 export { buildSkillActivationBlock, buildSkillDiscoveryVars };
@@ -292,10 +297,12 @@ function prependContextModeToBlock(
   block: string,
   renderMode: ContextModeRenderMode = "standalone",
 ): string {
+  const toolSurface = composeToolSurfaceInstructions(unitType, { renderMode });
   const contextMode = renderContextModeBlockForPrompt(unitType, base, renderMode);
-  if (!contextMode) return block;
-  if (!block.trim()) return contextMode;
-  return `${contextMode}\n\n${block}`;
+  const guidance = [toolSurface, contextMode].filter(Boolean).join("\n\n");
+  if (!guidance) return block;
+  if (!block.trim()) return guidance;
+  return `${guidance}\n\n${block}`;
 }
 
 function requireUnitPromptContextContract(unitType: string): UnitPromptContextContract {
@@ -969,7 +976,7 @@ export async function inlineDecisionsFromDb(
 
 /**
  * Inline requirements with optional milestone and slice scoping from the DB.
- * Falls back to filesystem via inlineGsdRootFile when DB unavailable or empty.
+ * Falls back to filesystem via inlineGsdRootFile only when DB is unavailable.
  */
 export async function inlineRequirementsFromDb(
   base: string, milestoneId?: string, sliceId?: string, level?: InlineLevel,
@@ -979,14 +986,28 @@ export async function inlineRequirementsFromDb(
     const { isDbAvailable } = await import("./gsd-db.js");
     if (isDbAvailable()) {
       const { queryRequirements, formatRequirementsForPrompt } = await import("./context-store.js");
-      const requirements = queryRequirements({ milestoneId, sliceId });
+      let requirements = queryRequirements({ milestoneId, sliceId });
+      let broadenedScope = false;
+      if (requirements.length === 0 && sliceId) {
+        requirements = queryRequirements({ milestoneId });
+        broadenedScope = true;
+      }
+      if (requirements.length === 0 && milestoneId) {
+        requirements = queryRequirements({ status: "active" });
+        broadenedScope = true;
+      }
       if (requirements.length > 0) {
-        // Use compact format for non-full levels to save ~40% tokens
-        const formatted = inlineLevel !== "full"
+        // Use compact format for non-full levels, milestone-scoped calls, and
+        // any cascade stage that broadened past the originally requested scope —
+        // a slice-scoped "full" call that fell through to the project-wide
+        // active fallback must not format those rows as full requirements.
+        const useCompact = inlineLevel !== "full" || !sliceId || broadenedScope;
+        const formatted = useCompact
           ? formatRequirementsCompact(requirements)
           : formatRequirementsForPrompt(requirements);
         return `### Requirements\nSource: \`.gsd/REQUIREMENTS.md\`\n\n${formatted}`;
       }
+      return null;
     }
   } catch (err) {
     logWarning("prompt", `inlineRequirementsFromDb failed: ${err instanceof Error ? err.message : String(err)}`);
@@ -1442,6 +1463,40 @@ export async function checkNeedsReassessment(
  * - No UAT file exists for the slice
  * - UAT result file already exists (idempotent — already ran)
  */
+/**
+ * Resolve the effective UAT mode for the dispatch gate the same way
+ * `buildRunUatPrompt` does: classify the UAT body with the slice's SUMMARY as
+ * supplemental context, so a `browser-executable` UAT whose slice references a
+ * self-contained harness (`npm run test:uat`, `search-uat.mjs`, `npx
+ * playwright test`) is promoted to `runtime-executable` for the gate too.
+ *
+ * Without this, `checkNeedsRunUat` returns `browser-executable` while the
+ * prompt instructs runtime-only execution, causing the dispatch gate to
+ * require browser tools / warm up the browser daemon (or stop dispatch when
+ * browser MCP is unavailable) for UAT runs that never touch the browser. See
+ * cursor[bot] review on PR #696 for the M007/S01 regression.
+ */
+async function resolveRunUatEffectiveType(
+  base: string,
+  mid: string,
+  sliceId: string,
+  uatContent: string,
+): Promise<UatType> {
+  let summaryContent = "";
+  try {
+    const summaryPath = resolveSliceFile(base, mid, sliceId, "SUMMARY");
+    if (summaryPath) {
+      summaryContent = (await loadFile(summaryPath)) ?? "";
+    }
+  } catch (err) {
+    logWarning(
+      "prompt",
+      `resolveRunUatEffectiveType SUMMARY load failed: ${err instanceof Error ? err.message : String(err)}`,
+    );
+  }
+  return classifyUatContentForRun(uatContent, summaryContent).effectiveType;
+}
+
 export async function checkNeedsRunUat(
   base: string, mid: string, state: GSDState, prefs: GSDPreferences | undefined,
 ): Promise<{ sliceId: string; uatType: UatType } | null> {
@@ -1470,7 +1525,8 @@ export async function checkNeedsRunUat(
             if (assessmentContent && hasVerdict(assessmentContent)) continue;
           }
           if (!shouldDispatchUatForContent(uatContent, prefs)) continue;
-          return { sliceId: sid, uatType: resolveEffectiveUatType(uatContent) };
+          const uatType = await resolveRunUatEffectiveType(base, mid, sid, uatContent);
+          return { sliceId: sid, uatType };
         }
         return null;
       }
@@ -1503,7 +1559,8 @@ export async function checkNeedsRunUat(
       if (assessmentContentFb && hasVerdict(assessmentContentFb)) continue;
     }
     if (!shouldDispatchUatForContent(uatContentFb, prefs)) continue;
-    return { sliceId: uatSid, uatType: resolveEffectiveUatType(uatContentFb) };
+    const uatType = await resolveRunUatEffectiveType(base, mid, uatSid, uatContentFb);
+    return { sliceId: uatSid, uatType };
   }
   return null;
 }
@@ -2470,6 +2527,9 @@ export async function buildPlanSlicePrompt(
       `Either (a) add an earlier task that creates X on disk before the task that needs it, ` +
       `or (b) if this task IS the one that creates X, move X from inputs to expected_output. ` +
       `Do NOT put X in a task's expected_output if that task only reads or verifies X — only tasks that actually write X to disk should list it in expected_output.\n` +
+      `- **"[file] X: ... GSD planning artifacts are projections preloaded as context / written by workflow tools"**: ` +
+      `Remove X from the task's inputs, files, and expectedOutput entirely. Planning artifacts (anything under .gsd/, .planning/, or .audits/, or names like M001-CONTEXT.md / S01-PLAN.md) are preloaded as context and written by workflow tools — ` +
+      `do NOT add a task that creates X and do NOT move X to expectedOutput.\n` +
       `- **"[file] X: Task T_early reads X but it's created by task T_late (sequence violation)"**: ` +
       `Either (a) reorder tasks so T_late (the creator) runs before T_early (the reader), ` +
       `or (b) if T_late doesn't actually create X (it only reads/tests it), remove X from T_late's expected_output entirely.\n` +
@@ -3554,15 +3614,30 @@ export async function buildRunUatPrompt(
     null,
     cappedInlinedContext.length < rawInlinedContext.length ? `dropped ${rawInlinedContext.length - cappedInlinedContext.length} chars` : "within budget",
   );
+  const uatPolicy = classifyUatContentForRun(uatContent, cappedInlinedContext);
+  const uatType = uatPolicy.effectiveType;
+  const runtimeHarnessOverride = uatPolicy.declaredType === "browser-executable" && uatType === "runtime-executable"
+    ? [
+      "## Runtime harness override",
+      "",
+      "This UAT declares `browser-executable` but the slice references a self-contained verification command (`npm run test:uat`, `search-uat.mjs`, or similar).",
+      "Run **only** that command via `gsd_uat_exec` with `uat-runtime-check` intent.",
+      "Do **not** call `uat-service-start`, do **not** run `npm run start`, `npm run test:server`, or any separate server command.",
+      "Do **not** use `browser_navigate` or other `browser_*` tools — the harness already exercises the browser.",
+      "When the harness exits 0, save **PASS** with `uatType: \"runtime-executable\"` (the effective mode above, not the UAT file header) and **runtime** check modes only.",
+      "",
+    ].join("\n")
+    : "";
   const inlinedContext = prependContextModeToBlock(
     "run-uat",
     base,
-    cappedInlinedContext,
+    runtimeHarnessOverride
+      ? `${runtimeHarnessOverride}\n${cappedInlinedContext}`
+      : cappedInlinedContext,
   );
   emitPromptContextTelemetry("run-uat", contextTelemetry, inlinedContext);
 
   const uatResultPath = join(base, relSliceFile(base, mid, sliceId, "ASSESSMENT"));
-  const uatType = resolveEffectiveUatType(uatContent);
   const canonicalPresentation = JSON.stringify(buildRunUatPresentationForType(uatType), null, 2);
 
   return loadPrompt("run-uat", {
@@ -3847,18 +3922,22 @@ export async function buildReactiveExecutePrompt(
   const inlinedTemplates = inlineTemplate("task-summary", "Task Summary");
   trackPromptContext(contextTelemetry, "templates", "inline", inlinedTemplates);
 
-  const prompt = loadPrompt("reactive-execute", {
-    workingDirectory: base,
-    milestoneId: mid,
-    milestoneTitle: midTitle,
-    sliceId: sid,
-    sliceTitle: sTitle,
-    graphContext: prependContextModeToBlock("reactive-execute", base, graphContext),
-    readyTaskCount: String(readyTaskIds.length),
-    readyTaskList: readyTaskListLines.join("\n"),
-    subagentPrompts: subagentSections.join("\n\n---\n\n"),
-    inlinedTemplates,
-  });
+  const prompt = prependContextModeToBlock(
+    "reactive-execute",
+    base,
+    loadPrompt("reactive-execute", {
+      workingDirectory: base,
+      milestoneId: mid,
+      milestoneTitle: midTitle,
+      sliceId: sid,
+      sliceTitle: sTitle,
+      graphContext,
+      readyTaskCount: String(readyTaskIds.length),
+      readyTaskList: readyTaskListLines.join("\n"),
+      subagentPrompts: subagentSections.join("\n\n---\n\n"),
+      inlinedTemplates,
+    }),
+  );
   emitPromptContextTelemetry("reactive-execute", contextTelemetry, prompt);
   return prompt;
 }
@@ -4037,17 +4116,21 @@ export async function buildGateEvaluatePrompt(
     ].join("\n"));
   }
 
-  return loadPrompt("gate-evaluate", {
-    workingDirectory: base,
-    milestoneId: mid,
-    milestoneTitle: midTitle,
-    sliceId: sid,
-    sliceTitle: sTitle,
-    slicePlanContent: prependContextModeToBlock("gate-evaluate", base, planContent),
-    gateCount: String(pending.length),
-    gateList: gateListLines.join("\n"),
-    subagentPrompts: subagentSections.join("\n\n---\n\n"),
-  });
+  return prependContextModeToBlock(
+    "gate-evaluate",
+    base,
+    loadPrompt("gate-evaluate", {
+      workingDirectory: base,
+      milestoneId: mid,
+      milestoneTitle: midTitle,
+      sliceId: sid,
+      sliceTitle: sTitle,
+      slicePlanContent: planContent,
+      gateCount: String(pending.length),
+      gateList: gateListLines.join("\n"),
+      subagentPrompts: subagentSections.join("\n\n---\n\n"),
+    }),
+  );
 }
 
 export async function buildRewriteDocsPrompt(

package/src/resources/extensions/gsd/auto-start.ts

@@ -16,6 +16,7 @@ import type {
   ExtensionCommandContext,
 } from "@gsd/pi-coding-agent";
 import { deriveState } from "./state.js";
+import { findWorktreeSegment, isGsdWorktreePath } from "./worktree-root.js";
 import { loadFile, getManifestStatus } from "./files.js";
 import type { InterruptedSessionAssessment } from "./interrupted-session.js";
 import {
@@ -27,6 +28,7 @@ import { ensureGsdSymlink, isInheritedRepo, validateProjectId } from "./repo-ide
 import { migrateToExternalState, recoverFailedMigration } from "./migrate-external.js";
 import { collectSecretsFromManifest } from "../get-secrets-from-user.js";
 import { gsdRoot, resolveMilestoneFile } from "./paths.js";
+import { milestoneEntryBlockedGuidance } from "./guidance.js";
 import { invalidateAllCaches } from "./cache.js";
 import { writeLock, clearLock, readCrashLock, isLockProcessAlive } from "./crash-recovery.js";
 import {
@@ -56,7 +58,8 @@ import {
   detectWorktreeName,
   setActiveMilestoneId,
 } from "./worktree.js";
-import { getAutoWorktreePath, isInAutoWorktree, checkoutBranchWithStashGuard } from "./auto-worktree.js";
+import { getAutoWorktreePath, isInAutoWorktree } from "./auto-worktree.js";
+import { checkoutBranchWithStashGuard } from "./worktree-git-recovery.js";
 import { readResourceVersion, cleanStaleRuntimeUnits } from "./auto-worktree.js";
 import { worktreePath as getWorktreeDir, isInsideWorktreesDir } from "./worktree-manager.js";
 import { emitWorktreeOrphaned } from "./worktree-telemetry.js";
@@ -96,7 +99,6 @@ import {
   rmSync,
 } from "node:fs";
 import { join } from "node:path";
-import { sep as pathSep } from "node:path";
 
 import { validateDirectory } from "./validate-directory.js";
 import {
@@ -601,7 +603,7 @@ export function auditOrphanedMilestoneBranches(
               warnings.push(`Failed to remove worktree directory for ${milestoneId}: ${err2 instanceof Error ? err2.message : String(err2)}`);
             }
           } else {
-            warnings.push(`Orphaned worktree directory for ${milestoneId} is outside .gsd/worktrees/ — skipping removal for safety.`);
+            warnings.push(`Orphaned worktree directory for ${milestoneId} is outside the GSD worktrees containers — skipping removal for safety.`);
           }
         } else {
           pushAction({
@@ -717,7 +719,7 @@ export function auditOrphanedMilestoneBranches(
     if (!existsSync(wtDir)) continue;
     if (!isInsideWorktreesDir(basePath, wtDir)) {
       warnings.push(
-        `Orphaned worktree directory for ${m.id} is outside .gsd/worktrees/ — skipping removal for safety.`,
+        `Orphaned worktree directory for ${m.id} is outside the GSD worktrees containers — skipping removal for safety.`,
       );
       continue;
     }
@@ -993,12 +995,14 @@ export async function bootstrapAutoSession(
   // phase-specific planning model for a discuss turn (#2829).
   //
   // Precedence:
-  // 1) Explicit session override via /gsd model (this session)
-  // 2) Current session model from settings/session restore (if provider ready)
-  // 3) GSD model preferences from PREFERENCES.md (validated against live auth)
+  // 1) Explicit session override via /gsd model or /gsd auto --model (this session)
+  // 2) GSD model preferences from PREFERENCES.md (validated against live auth)
+  // 3) Current session model from settings/session restore (if provider ready)
   //
-  // This preserves #3517 defaults while honoring explicit runtime model
-  // selection for subsequent /gsd runs in the same session.
+  // PREFERENCES.md wins over the ambient session default (#3517) so /gsd auto
+  // does not stick on claude-code/claude-sonnet-4-6 when the user configured
+  // models via /gsd workflow-preferences or PREFERENCES.md. Custom providers
+  // still skip PREFERENCES.md entirely (#4122).
   //
   // Exception (#4122): when the session provider is a custom provider declared
   // in ~/.gsd/agent/models.json (Ollama, vLLM, OpenAI-compatible proxy, etc.),
@@ -1010,7 +1014,7 @@ export async function bootstrapAutoSession(
   const sessionProviderIsCustom = isCustomProvider(ctx.model?.provider);
   const preferredModel = sessionProviderIsCustom
     ? null
-    : resolveDefaultSessionModel(ctx.model?.provider);
+    : resolveDefaultSessionModel(ctx.model?.provider, base);
   // Validate the preferred model against the live registry + provider auth so
   // an unconfigured PREFERENCES.md entry (no API key / OAuth) can't become the
   // start-model snapshot. Without this, every subsequent unit would try to
@@ -1040,8 +1044,8 @@ export async function bootstrapAutoSession(
     : null;
   const startThinkingSnapshot = pi.getThinkingLevel();
   const startModelSnapshot = manualSessionOverride
-    ?? currentSessionModel
     ?? validatedPreferredModel
+    ?? currentSessionModel
     ?? null;
 
   try {
@@ -1267,11 +1271,29 @@ export async function bootstrapAutoSession(
       }
     }
 
-    const blockingStrandedRecoveryAction = state.activeMilestone
-      ? strandedRecoveryActions.find(
+    const requestedMilestoneLock = process.env.GSD_MILESTONE_LOCK?.trim() || null;
+    const lockedActiveMilestone =
+      requestedMilestoneLock && state.activeMilestone?.id === requestedMilestoneLock;
+    let blockingStrandedRecoveryAction: OrphanAuditAction | null;
+    if (lockedActiveMilestone) {
+      // Parallel worker or explicit `/gsd auto Mxxx`: sibling milestones'
+      // stranded work must not block this milestone's resumption, and the
+      // downstream `strandedRecoveryAction` (used for currentMilestoneId,
+      // setActiveMilestoneId, and adoptStrandedMilestone) must be scoped to
+      // the locked milestone only. Falling back to the first sibling action
+      // would mis-target adoption (#742).
+      const lockMatch = strandedRecoveryActions.find(
+        (action) => action.milestoneId === requestedMilestoneLock,
+      ) ?? null;
+      blockingStrandedRecoveryAction = lockMatch;
+      strandedRecoveryAction = lockMatch;
+    } else if (state.activeMilestone) {
+      blockingStrandedRecoveryAction = strandedRecoveryActions.find(
         (action) => action.milestoneId !== state.activeMilestone?.id,
-      ) ?? strandedRecoveryAction
-      : strandedRecoveryAction;
+      ) ?? strandedRecoveryAction;
+    } else {
+      blockingStrandedRecoveryAction = strandedRecoveryAction;
+    }
 
     if (blockingStrandedRecoveryAction) {
       if (!state.activeMilestone) {
@@ -1293,6 +1315,8 @@ export async function bootstrapAutoSession(
         formatStrandedWorkRecoveryMessage(strandedRecoveryAction),
         "info",
       );
+    } else if (lockedActiveMilestone) {
+      strandedRecoveryAction = null;
     }
 
     if (
@@ -1325,7 +1349,7 @@ export async function bootstrapAutoSession(
       (state.phase === "pre-planning" || state.phase === "complete") &&
       survivorIsolationMode !== "none" &&
       !detectWorktreeName(base) &&
-      !base.includes(`${pathSep}.gsd${pathSep}worktrees${pathSep}`)
+      !isGsdWorktreePath(base)
     ) {
       const milestoneBranch = `milestone/${survivorMilestoneId}`;
       const { nativeBranchExists } = await import("./native-git-bridge.js");
@@ -1610,16 +1634,12 @@ export async function bootstrapAutoSession(
     // live here is gone.
 
     const isUnderGsdWorktrees = (p: string): boolean => {
-      // Direct layout: /.gsd/worktrees/
-      const marker = `${pathSep}.gsd${pathSep}worktrees${pathSep}`;
-      if (p.includes(marker)) return true;
-      const worktreesSuffix = `${pathSep}.gsd${pathSep}worktrees`;
-      if (p.endsWith(worktreesSuffix)) return true;
-      // Symlink-resolved layout: /.gsd/projects/<hash>/worktrees/
-      const symlinkRe = new RegExp(
-        `\\${pathSep}\\.gsd\\${pathSep}projects\\${pathSep}[a-f0-9]+\\${pathSep}worktrees(?:\\${pathSep}|$)`,
-      );
-      return symlinkRe.test(p);
+      const normalized = p.replaceAll("\\", "/");
+      if (findWorktreeSegment(normalized) !== null) return true;
+      // The container directory itself (no trailing worktree name), in any layout.
+      return normalized.endsWith("/.gsd/worktrees")
+        || normalized.endsWith("/.gsd-worktrees")
+        || /\/\.gsd\/projects\/[^/]+\/worktrees$/.test(normalized);
     };
 
     if (
@@ -1646,14 +1666,9 @@ export async function bootstrapAutoSession(
             `Cannot enter milestone ${s.currentMilestoneId}: lease is held by another worker.`,
             "error",
           );
-        } else if (enterResult.reason === "creation-failed") {
-          ctx.ui.notify(
-            `Cannot enter milestone ${s.currentMilestoneId}: worktree/branch creation failed. Isolation is degraded.`,
-            "error",
-          );
-        } else if (enterResult.reason === "isolation-degraded") {
+        } else if (enterResult.reason === "creation-failed" || enterResult.reason === "isolation-degraded") {
           ctx.ui.notify(
-            `Cannot enter milestone ${s.currentMilestoneId}: isolation is degraded from a prior worktree failure. Close processes locking the worktree and retry, or run /gsd doctor fix.`,
+            milestoneEntryBlockedGuidance(s.currentMilestoneId, enterResult.reason),
             "error",
           );
         } else if (enterResult.reason === "invalid-milestone-id") {

package/src/resources/extensions/gsd/auto-tool-tracking.ts

@@ -5,6 +5,7 @@
  */
 
 import { stripMcpToolPrefix } from "@gsd/pi-ai";
+import { TOOL_SURFACE_NOT_READY } from "./tool-surface-readiness.js";
 
 interface InFlightTool {
   startedAt: number;
@@ -132,6 +133,15 @@ export function clearInFlightTools(): void {
 const TOOL_INVOCATION_ERROR_RE = /Validation failed for tool|Input validation error|Invalid arguments for tool|MCP error -32602|No such tool available|Expected ',' or '\}'(?: after property value)?(?: in JSON)?|Unexpected end of JSON|Unexpected token.*in JSON|does not provide an export named|Named export .* not found|Cannot find module|ERR_MODULE_NOT_FOUND|ERR_MODULE_NOT_EXPORTED|ERR_PACKAGE_PATH_NOT_EXPORTED/i;
 const DETERMINISTIC_POLICY_ERROR_RE = /(?:^|\b)(?:HARD BLOCK:|Blocked: \/gsd queue is a planning tool|Direct writes to \.gsd\/STATE\.md and \.gsd\/gsd\.db are blocked|This is a mechanical gate)/i;
 
+/**
+ * Matches the runtime's "tool not registered" error. Unlike the deterministic
+ * invocation failures above, this one is usually transient: the workflow MCP
+ * server registers its tool surface asynchronously after session start, so a
+ * Unit's first tool call can race the registration. Callers should retry
+ * (bounded) instead of breaking the loop.
+ */
+const TOOL_UNAVAILABLE_ERROR_RE = new RegExp(`No such tool available|${TOOL_SURFACE_NOT_READY}`, "i");
+
 /**
  * Returns true if the error message indicates a deterministic invocation or
  * policy failure (as opposed to a normal tool execution error).
@@ -141,6 +151,15 @@ export function isToolInvocationError(errorMsg: string): boolean {
   return TOOL_INVOCATION_ERROR_RE.test(errorMsg) || isDeterministicPolicyError(errorMsg);
 }
 
+/**
+ * Returns true if the error message indicates the called tool was not on the
+ * session's tool surface (MCP startup race — see TOOL_UNAVAILABLE_ERROR_RE).
+ */
+export function isToolUnavailableError(errorMsg: string): boolean {
+  if (!errorMsg) return false;
+  return TOOL_UNAVAILABLE_ERROR_RE.test(errorMsg);
+}
+
 /**
  * Returns true if the error message indicates the tool was skipped because
  * a queued user message interrupted the turn (#3595).  Retrying will produce