gsd-pi 2.76.0-dev.4100bd590 → 2.76.0-dev.82e249f7b

package/scripts/link-workspace-packages.cjs

@@ -34,6 +34,7 @@ const packageMap = {
   'pi-coding-agent': { scope: '@gsd', name: 'pi-coding-agent' },
   'pi-tui': { scope: '@gsd', name: 'pi-tui' },
   'rpc-client': { scope: '@gsd-build', name: 'rpc-client' },
+  'mcp-server': { scope: '@gsd-build', name: 'mcp-server' },
 }
 
 for (const scopeDir of Object.values(scopeDirs)) {

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

@@ -213,6 +213,7 @@ export interface LoopDeps {
     retryContext?: { isRetry: boolean; previousTier?: string },
     isAutoMode?: boolean,
     sessionModelOverride?: { provider: string; id: string } | null,
+    autoModeStartThinkingLevel?: ReturnType<ExtensionAPI["getThinkingLevel"]> | null,
   ) => Promise<{
     routing: { tier: string; modelDowngraded: boolean } | null;
     appliedModel: { provider: string; id: string } | null;

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

@@ -1471,6 +1471,7 @@ export async function runUnitPhase(
     sidecarItem ? undefined : { isRetry, previousTier },
     undefined,
     s.manualSessionModelOverride,
+    s.autoModeStartThinkingLevel,
   );
   s.currentUnitRouting =
     modelResult.routing as AutoSession["currentUnitRouting"];
@@ -1485,6 +1486,9 @@ export async function runUnitPhase(
     if (match) {
       const ok = await pi.setModel(match, { persist: false });
       if (ok) {
+        if (s.autoModeStartThinkingLevel) {
+          pi.setThinkingLevel(s.autoModeStartThinkingLevel);
+        }
         s.currentUnitModel = match as AutoSession["currentUnitModel"];
         ctx.ui.notify(`Hook model override: ${match.provider}/${match.id}`, "info");
       } else {

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

@@ -17,7 +17,7 @@
  */
 
 import type { Api, Model } from "@gsd/pi-ai";
-import type { ExtensionCommandContext } from "@gsd/pi-coding-agent";
+import type { ExtensionAPI, ExtensionCommandContext } from "@gsd/pi-coding-agent";
 import type { GitServiceImpl } from "../git-service.js";
 import type { CaptureEntry } from "../captures.js";
 import type { BudgetAlertLevel } from "../auto-budget.js";
@@ -40,6 +40,8 @@ export interface StartModel {
   id: string;
 }
 
+export type ThinkingLevelSnapshot = ReturnType<ExtensionAPI["getThinkingLevel"]>;
+
 export interface PendingVerificationRetry {
   unitId: string;
   failureContext: string;
@@ -120,6 +122,8 @@ export class AutoSession {
   currentDispatchedModelId: string | null = null;
   originalModelId: string | null = null;
   originalModelProvider: string | null = null;
+  autoModeStartThinkingLevel: ThinkingLevelSnapshot | null = null;
+  originalThinkingLevel: ThinkingLevelSnapshot | null = null;
   lastBudgetAlertLevel: BudgetAlertLevel = 0;
 
   // ── Recovery ─────────────────────────────────────────────────────────────
@@ -241,6 +245,8 @@ export class AutoSession {
     this.currentDispatchedModelId = null;
     this.originalModelId = null;
     this.originalModelProvider = null;
+    this.autoModeStartThinkingLevel = null;
+    this.originalThinkingLevel = null;
     this.lastBudgetAlertLevel = 0;
 
     // Recovery

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

@@ -33,6 +33,14 @@ export interface PreferredModelConfig {
   source: "explicit" | "synthesized";
 }
 
+function reapplyThinkingLevel(
+  pi: ExtensionAPI,
+  level: ReturnType<ExtensionAPI["getThinkingLevel"]> | null | undefined,
+): void {
+  if (!level) return;
+  pi.setThinkingLevel(level);
+}
+
 export function resolvePreferredModelConfig(
   unitType: string,
   autoModeStartModel: { provider: string; id: string; flatRateCtx?: FlatRateContext } | null,
@@ -97,6 +105,8 @@ export async function selectAndApplyModel(
   isAutoMode = true,
   /** Explicit /gsd model pin captured at bootstrap for long-running auto loops. */
   sessionModelOverride?: { provider: string; id: string } | null,
+  /** Thinking level captured at auto-mode start and re-applied after model swaps. */
+  autoModeStartThinkingLevel?: ReturnType<ExtensionAPI["getThinkingLevel"]> | null,
 ): Promise<ModelSelectionResult> {
   const uokFlags = resolveUokFlags(prefs);
   const effectiveSessionModelOverride = sessionModelOverride === undefined
@@ -380,6 +390,7 @@ export async function selectAndApplyModel(
       const ok = await pi.setModel(model, { persist: false });
       if (ok) {
         appliedModel = model;
+        reapplyThinkingLevel(pi, autoModeStartThinkingLevel);
 
         // ADR-005: Adjust active tool set for the selected model's provider capabilities.
         // Hard-filter incompatible tools, then let extensions override via adjust_tool_set hook.
@@ -456,10 +467,14 @@ export async function selectAndApplyModel(
           );
           if (byId) {
             const fallbackOk = await pi.setModel(byId, { persist: false });
-            if (fallbackOk) appliedModel = byId;
+            if (fallbackOk) {
+              appliedModel = byId;
+              reapplyThinkingLevel(pi, autoModeStartThinkingLevel);
+            }
           }
         } else {
           appliedModel = startModel;
+          reapplyThinkingLevel(pi, autoModeStartThinkingLevel);
         }
       }
     }

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

@@ -273,8 +273,8 @@ export async function bootstrapAutoSession(
   //
   // Precedence:
   // 1) Explicit session override via /gsd 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)
+  // 2) Current session model from settings/session restore (if provider ready)
+  // 3) GSD model preferences from PREFERENCES.md (validated against live auth)
   //
   // This preserves #3517 defaults while honoring explicit runtime model
   // selection for subsequent /gsd runs in the same session.
@@ -314,11 +314,14 @@ export async function bootstrapAutoSession(
   }
   const sessionModelReady =
     ctx.model && ctx.modelRegistry.isProviderRequestReady(ctx.model.provider);
+  const currentSessionModel = (sessionModelReady && ctx.model)
+    ? { provider: ctx.model.provider, id: ctx.model.id }
+    : null;
+  const startThinkingSnapshot = pi.getThinkingLevel();
   const startModelSnapshot = manualSessionOverride
+    ?? currentSessionModel
     ?? validatedPreferredModel
-    ?? (sessionModelReady && ctx.model
-      ? { provider: ctx.model.provider, id: ctx.model.id }
-      : null);
+    ?? null;
 
   try {
     // Validate GSD_PROJECT_ID early so the user gets immediate feedback
@@ -664,8 +667,9 @@ export async function bootstrapAutoSession(
     s.pendingQuickTasks = [];
     s.currentUnit = null;
     s.currentMilestoneId = state.activeMilestone?.id ?? null;
-    s.originalModelId = ctx.model?.id ?? null;
-    s.originalModelProvider = ctx.model?.provider ?? null;
+    s.originalModelId = startModelSnapshot?.id ?? ctx.model?.id ?? null;
+    s.originalModelProvider = startModelSnapshot?.provider ?? ctx.model?.provider ?? null;
+    s.originalThinkingLevel = startThinkingSnapshot ?? null;
 
     // Register SIGTERM handler
     registerSigtermHandler(base);
@@ -779,6 +783,7 @@ export async function bootstrapAutoSession(
         id: startModelSnapshot.id,
       };
     }
+    s.autoModeStartThinkingLevel = startThinkingSnapshot ?? null;
     s.manualSessionModelOverride = manualSessionOverride ?? null;
 
     // Apply worker model override from parallel orchestrator (#worker-model).

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

@@ -969,7 +969,7 @@ export async function stopAuto(
       logWarning("engine", `file unlink failed: ${err instanceof Error ? err.message : String(err)}`, { file: "auto.ts" });
     }
 
-    // ── Step 13: Restore original model (before reset clears IDs) ──
+    // ── Step 13: Restore original model + thinking (before reset clears IDs) ──
     try {
       if (pi && ctx && s.originalModelId && s.originalModelProvider) {
         const original = ctx.modelRegistry.find(
@@ -978,6 +978,9 @@ export async function stopAuto(
         );
         if (original) await pi.setModel(original);
       }
+      if (pi && s.originalThinkingLevel) {
+        pi.setThinkingLevel(s.originalThinkingLevel);
+      }
     } catch (e) {
       debugLog("stop-cleanup-model", { error: e instanceof Error ? e.message : String(e) });
     }

package/src/resources/extensions/gsd/complexity-classifier.ts

@@ -32,11 +32,13 @@ export interface TaskMetadata {
 // ─── Unit Type → Default Tier Mapping ────────────────────────────────────────
 
 const UNIT_TYPE_TIERS: Record<string, ComplexityTier> = {
-  // Tier 1 — Light: structured summaries, completion, UAT
-  "complete-slice": "light",
+  // Tier 1 — Light: compact verification turns
   "run-uat": "light",
 
-  // Tier 2 — Standard: research, routine discussion
+  // Tier 2 — Standard: research, routine discussion, slice completion
+  // complete-slice can carry large inlined context; avoid routing it to the
+  // cheapest "light" model by default (#4520).
+  "complete-slice": "standard",
   "discuss-milestone": "standard",
   "discuss-slice": "standard",
   "research-milestone": "standard",

package/src/resources/extensions/gsd/prompt-loader.ts

@@ -24,6 +24,35 @@ import { fileURLToPath } from "node:url";
 import { homedir } from "node:os";
 import { logWarning } from "./workflow-logger.js";
 
+type ExistsFn = (path: string) => boolean;
+
+function hasRequiredExtensionAssets(rootDir: string, exists: ExistsFn = existsSync): boolean {
+  return (
+    exists(join(rootDir, "prompts")) &&
+    exists(join(rootDir, "templates", "task-summary.md"))
+  );
+}
+
+export function resolveExtensionDirFromCandidates(
+  moduleDir: string,
+  agentGsdDir: string,
+  exists: ExistsFn = existsSync,
+): string {
+  const moduleUsable = hasRequiredExtensionAssets(moduleDir, exists);
+  const agentUsable = hasRequiredExtensionAssets(agentGsdDir, exists);
+
+  // Prefer the user-local extension tree when both are valid. This avoids
+  // leaking npm/global-install paths into prompts on Windows.
+  if (agentUsable) return agentGsdDir;
+  if (moduleUsable) return moduleDir;
+
+  // Degraded fallback: if required template is missing in both locations,
+  // keep previous behavior and prefer whichever still has prompts/.
+  if (exists(join(moduleDir, "prompts"))) return moduleDir;
+  if (exists(join(agentGsdDir, "prompts"))) return agentGsdDir;
+  return moduleDir;
+}
+
 /**
  * Resolve the GSD extension directory.
  *
@@ -36,15 +65,9 @@ import { logWarning } from "./workflow-logger.js";
  */
 function resolveExtensionDir(): string {
   const moduleDir = dirname(fileURLToPath(import.meta.url));
-  if (existsSync(join(moduleDir, "prompts"))) return moduleDir;
-
-  // Fallback: user-local agent directory
   const gsdHome = process.env.GSD_HOME || join(homedir(), ".gsd");
   const agentGsdDir = join(gsdHome, "agent", "extensions", "gsd");
-  if (existsSync(join(agentGsdDir, "prompts"))) return agentGsdDir;
-
-  // Last resort: return the module dir (warmCache will silently handle the miss)
-  return moduleDir;
+  return resolveExtensionDirFromCandidates(moduleDir, agentGsdDir);
 }
 
 const __extensionDir = resolveExtensionDir();

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

@@ -341,6 +341,18 @@ test("dynamic routing passes provider-qualified model keys to the router", () =>
   );
 });
 
+test("selectAndApplyModel re-applies captured thinking level after setModel success", () => {
+  const src = readFileSync(join(__dirname, "..", "auto-model-selection.ts"), "utf-8");
+  assert.ok(
+    src.includes("autoModeStartThinkingLevel?: ReturnType<ExtensionAPI[\"getThinkingLevel\"]> | null"),
+    "selectAndApplyModel should accept an autoModeStartThinkingLevel parameter",
+  );
+  assert.ok(
+    src.includes("reapplyThinkingLevel(pi, autoModeStartThinkingLevel)"),
+    "selectAndApplyModel should re-apply captured thinking level after model changes",
+  );
+});
+
 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-start-model-capture.test.ts

@@ -52,9 +52,7 @@ test("bootstrapAutoSession checks manual session override before preferences", (
     "manual override and preference fallback must be resolved before building startModelSnapshot",
   );
 
-  // The validated preferred model must still appear as one of the snapshot
-  // sources so PREFERENCES.md continues to win over a stale settings.json
-  // default for built-in providers.
+  // Preferred model should still be part of fallback resolution.
   const snapshotBlock = source.slice(snapshotIdx, snapshotIdx + 400);
   assert.ok(
     snapshotBlock.includes("validatedPreferredModel") || snapshotBlock.includes("preferredModel"),
@@ -62,6 +60,22 @@ test("bootstrapAutoSession checks manual session override before preferences", (
   );
 });
 
+test("bootstrapAutoSession prioritizes current session model over PREFERENCES.md default", () => {
+  const snapshotIdx = source.indexOf("const startModelSnapshot = manualSessionOverride");
+  assert.ok(snapshotIdx > -1, "auto-start.ts should build startModelSnapshot");
+
+  const snapshotBlock = source.slice(snapshotIdx, snapshotIdx + 500);
+  const currentIdx = snapshotBlock.indexOf("currentSessionModel");
+  const preferredIdx = snapshotBlock.indexOf("validatedPreferredModel");
+
+  assert.ok(currentIdx > -1, "startModelSnapshot should include currentSessionModel");
+  assert.ok(preferredIdx > -1, "startModelSnapshot should include validatedPreferredModel");
+  assert.ok(
+    currentIdx < preferredIdx,
+    "startModelSnapshot should prefer currentSessionModel before validatedPreferredModel",
+  );
+});
+
 test("bootstrapAutoSession prefers session model over PREFERENCES.md when provider is custom (#4122)", () => {
   // Custom providers (Ollama, vLLM, OpenAI-compatible proxies) live in
   // ~/.gsd/agent/models.json, not PREFERENCES.md.  When the user picks one
@@ -111,3 +125,19 @@ test("bootstrapAutoSession validates preferred model against live registry auth
   const warningIdx = source.indexOf("is not configured; falling back to session default");
   assert.ok(warningIdx > -1, "auto-start.ts should warn when preferred model is unconfigured");
 });
+
+test("bootstrapAutoSession snapshots and persists thinking level for auto-mode lifecycle", () => {
+  const captureIdx = source.indexOf("const startThinkingSnapshot = pi.getThinkingLevel()");
+  assert.ok(captureIdx > -1, "auto-start.ts should snapshot thinking level at bootstrap start");
+
+  const originalThinkingIdx = source.indexOf("s.originalThinkingLevel = startThinkingSnapshot ?? null");
+  assert.ok(originalThinkingIdx > -1, "auto-start.ts should store originalThinkingLevel from snapshot");
+
+  const autoThinkingIdx = source.indexOf("s.autoModeStartThinkingLevel = startThinkingSnapshot ?? null");
+  assert.ok(autoThinkingIdx > -1, "auto-start.ts should store autoModeStartThinkingLevel from snapshot");
+
+  assert.ok(
+    captureIdx < originalThinkingIdx && captureIdx < autoThinkingIdx,
+    "thinking snapshot must be captured before session state assignment",
+  );
+});

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

@@ -0,0 +1,38 @@
+import test from "node:test";
+import assert from "node:assert/strict";
+import { readFileSync } from "node:fs";
+import { join } from "node:path";
+
+const autoSrc = readFileSync(join(import.meta.dirname, "..", "auto.ts"), "utf-8");
+const phasesSrc = readFileSync(join(import.meta.dirname, "..", "auto", "phases.ts"), "utf-8");
+
+test("stopAuto restores original thinking level", () => {
+  assert.ok(
+    autoSrc.includes("if (pi && s.originalThinkingLevel)"),
+    "auto.ts should conditionally restore original thinking level in stopAuto",
+  );
+  assert.ok(
+    autoSrc.includes("pi.setThinkingLevel(s.originalThinkingLevel)"),
+    "auto.ts should call pi.setThinkingLevel with originalThinkingLevel",
+  );
+});
+
+test("runUnitPhase threads captured thinking level into selectAndApplyModel", () => {
+  const callIdx = phasesSrc.indexOf("deps.selectAndApplyModel(");
+  assert.ok(callIdx > -1, "phases.ts should call selectAndApplyModel");
+  const callBlock = phasesSrc.slice(callIdx, callIdx + 600);
+  assert.ok(
+    callBlock.includes("s.autoModeStartThinkingLevel"),
+    "runUnitPhase should pass autoModeStartThinkingLevel to selectAndApplyModel",
+  );
+});
+
+test("hook model override preserves captured thinking level", () => {
+  const hookIdx = phasesSrc.indexOf("const hookModelOverride = sidecarItem?.model ?? iterData.hookModelOverride;");
+  assert.ok(hookIdx > -1, "phases.ts should include hook model override handling");
+  const hookBlock = phasesSrc.slice(hookIdx, hookIdx + 600);
+  assert.ok(
+    hookBlock.includes("pi.setThinkingLevel(s.autoModeStartThinkingLevel)"),
+    "hook model override should re-apply captured thinking level after setModel",
+  );
+});

package/src/resources/extensions/gsd/tests/complexity-classifier.test.ts

@@ -21,9 +21,9 @@ test("tierOrdinal returns correct ordering", () => {
 
 // ─── Unit Type Classification ────────────────────────────────────────────────
 
-test("complete-slice classifies as light", () => {
+test("complete-slice classifies as standard", () => {
   const result = classifyUnitComplexity("complete-slice", "M001/S01", "/tmp/fake");
-  assert.equal(result.tier, "light");
+  assert.equal(result.tier, "standard");
 });
 
 test("run-uat classifies as light", () => {
@@ -145,7 +145,7 @@ test("budget pressure at 90% downgrades standard to light", () => {
   assert.equal(result.downgraded, true);
 });
 
-test("budget pressure at 90% downgrades light stays light", () => {
+test("budget pressure at 90% downgrades complete-slice standard to light", () => {
   const result = classifyUnitComplexity("complete-slice", "M001/S01", "/tmp/fake", 0.95);
   assert.equal(result.tier, "light");
 });

package/src/resources/extensions/gsd/tests/prompt-loader-extension-dir.test.ts

@@ -0,0 +1,49 @@
+import test from "node:test";
+import assert from "node:assert/strict";
+import { join } from "node:path";
+
+import { resolveExtensionDirFromCandidates } from "../prompt-loader.ts";
+
+function makeExists(paths: Set<string>): (path: string) => boolean {
+  return (path: string) => paths.has(path);
+}
+
+test("resolveExtensionDirFromCandidates prefers user-local dir when both trees are valid", () => {
+  const moduleDir = "/npm/global/gsd";
+  const agentDir = "/home/user/.gsd/agent/extensions/gsd";
+  const paths = new Set<string>([
+    join(moduleDir, "prompts"),
+    join(moduleDir, "templates", "task-summary.md"),
+    join(agentDir, "prompts"),
+    join(agentDir, "templates", "task-summary.md"),
+  ]);
+
+  const resolved = resolveExtensionDirFromCandidates(moduleDir, agentDir, makeExists(paths));
+  assert.equal(resolved, agentDir);
+});
+
+test("resolveExtensionDirFromCandidates rejects module dir missing task-summary template", () => {
+  const moduleDir = "/npm/global/gsd";
+  const agentDir = "/home/user/.gsd/agent/extensions/gsd";
+  const paths = new Set<string>([
+    join(moduleDir, "prompts"),
+    // Missing module templates/task-summary.md on purpose.
+    join(agentDir, "prompts"),
+    join(agentDir, "templates", "task-summary.md"),
+  ]);
+
+  const resolved = resolveExtensionDirFromCandidates(moduleDir, agentDir, makeExists(paths));
+  assert.equal(resolved, agentDir);
+});
+
+test("resolveExtensionDirFromCandidates falls back to prompts-only dir when neither tree is fully valid", () => {
+  const moduleDir = "/npm/global/gsd";
+  const agentDir = "/home/user/.gsd/agent/extensions/gsd";
+  const paths = new Set<string>([
+    join(moduleDir, "prompts"),
+    // Neither side has templates/task-summary.md.
+  ]);
+
+  const resolved = resolveExtensionDirFromCandidates(moduleDir, agentDir, makeExists(paths));
+  assert.equal(resolved, moduleDir);
+});