gsd-pi 2.37.1-dev.d3ace49 → 2.38.0-dev.63ad7e5

package/src/resources/extensions/gsd/tests/cmux.test.ts

@@ -100,6 +100,99 @@ test("buildCmuxStatusLabel and progress prefer deepest active unit", () => {
   assert.deepEqual(buildCmuxProgress(state), { value: 0.4, label: "2/5 tasks" });
 });
 
+describe("createGridLayout", () => {
+  // Create a mock CmuxClient that tracks createSplitFrom calls
+  function makeMockClient() {
+    let nextId = 1;
+    const calls: Array<{ source: string | undefined; direction: string }> = [];
+
+    const client = {
+      calls,
+      async createGridLayout(count: number) {
+        // Simulate the grid layout logic with a fake client
+        if (count <= 0) return [];
+        const surfaces: string[] = [];
+
+        const createSplitFrom = async (source: string | undefined, direction: string) => {
+          calls.push({ source, direction });
+          return `surface-${nextId++}`;
+        };
+
+        const rightCol = await createSplitFrom("gsd-surface", "right");
+        surfaces.push(rightCol);
+        if (count === 1) return surfaces;
+
+        const bottomRight = await createSplitFrom(rightCol, "down");
+        surfaces.push(bottomRight);
+        if (count === 2) return surfaces;
+
+        const bottomLeft = await createSplitFrom("gsd-surface", "down");
+        surfaces.push(bottomLeft);
+        if (count === 3) return surfaces;
+
+        let lastSurface = bottomRight;
+        for (let i = 3; i < count; i++) {
+          const next = await createSplitFrom(lastSurface, "down");
+          surfaces.push(next);
+          lastSurface = next;
+        }
+
+        return surfaces;
+      },
+    };
+    return client;
+  }
+
+  test("1 agent creates single right split", async () => {
+    const mock = makeMockClient();
+    const surfaces = await mock.createGridLayout(1);
+    assert.equal(surfaces.length, 1);
+    assert.deepEqual(mock.calls, [
+      { source: "gsd-surface", direction: "right" },
+    ]);
+  });
+
+  test("2 agents creates right column then splits it down", async () => {
+    const mock = makeMockClient();
+    const surfaces = await mock.createGridLayout(2);
+    assert.equal(surfaces.length, 2);
+    assert.deepEqual(mock.calls, [
+      { source: "gsd-surface", direction: "right" },
+      { source: "surface-1", direction: "down" },
+    ]);
+  });
+
+  test("3 agents creates 2x2 grid (gsd + 3 agent surfaces)", async () => {
+    const mock = makeMockClient();
+    const surfaces = await mock.createGridLayout(3);
+    assert.equal(surfaces.length, 3);
+    assert.deepEqual(mock.calls, [
+      { source: "gsd-surface", direction: "right" },
+      { source: "surface-1", direction: "down" },
+      { source: "gsd-surface", direction: "down" },
+    ]);
+  });
+
+  test("4 agents creates 2x2 grid with extra split", async () => {
+    const mock = makeMockClient();
+    const surfaces = await mock.createGridLayout(4);
+    assert.equal(surfaces.length, 4);
+    assert.deepEqual(mock.calls, [
+      { source: "gsd-surface", direction: "right" },
+      { source: "surface-1", direction: "down" },
+      { source: "gsd-surface", direction: "down" },
+      { source: "surface-2", direction: "down" },
+    ]);
+  });
+
+  test("0 agents returns empty", async () => {
+    const mock = makeMockClient();
+    const surfaces = await mock.createGridLayout(0);
+    assert.equal(surfaces.length, 0);
+    assert.equal(mock.calls.length, 0);
+  });
+});
+
 describe("cmux extension discovery opt-out", () => {
   test("cmux directory has package.json with pi manifest to prevent auto-discovery as extension", () => {
     const cmuxDir = path.resolve(

package/src/resources/extensions/gsd/tests/doctor-enhancements.test.ts

@@ -0,0 +1,266 @@
+import { mkdtempSync, mkdirSync, rmSync, writeFileSync, existsSync } from "node:fs";
+import { join } from "node:path";
+import { tmpdir } from "node:os";
+
+import { runGSDDoctor } from "../doctor.js";
+import { formatDoctorReportJson } from "../doctor-format.js";
+import { createTestContext } from "./test-helpers.ts";
+
+const { assertEq, assertTrue, assertMatch, report } = createTestContext();
+
+// ── Helpers ─────────────────────────────────────────────────────────────────
+
+function makeBase(): { base: string; gsd: string; mDir: string } {
+  const base = mkdtempSync(join(tmpdir(), "gsd-doctor-enh-"));
+  const gsd = join(base, ".gsd");
+  const mDir = join(gsd, "milestones", "M001");
+  mkdirSync(join(mDir, "slices"), { recursive: true });
+  return { base, gsd, mDir };
+}
+
+function writeRoadmap(mDir: string, content: string): void {
+  writeFileSync(join(mDir, "M001-ROADMAP.md"), content);
+}
+
+function writeSlice(mDir: string, sliceId: string, planContent: string): string {
+  const sDir = join(mDir, "slices", sliceId);
+  const tDir = join(sDir, "tasks");
+  mkdirSync(tDir, { recursive: true });
+  writeFileSync(join(sDir, `${sliceId}-PLAN.md`), planContent);
+  return sDir;
+}
+
+async function main(): Promise<void> {
+  // ── 1. Circular dependency detection ──────────────────────────────────────
+  console.log("\n=== circular dependency detection ===");
+  {
+    const { base, mDir } = makeBase();
+    writeRoadmap(mDir, `# M001: Circular Test\n\n## Slices\n- [ ] **S01: Slice A** \`risk:low\` \`depends:[S02]\`\n  > After this: done\n- [ ] **S02: Slice B** \`risk:low\` \`depends:[S01]\`\n  > After this: done\n`);
+    writeSlice(mDir, "S01", "# S01: Slice A\n\n**Goal:** A\n**Demo:** A\n\n## Tasks\n- [ ] **T01: Task** `est:10m`\n  Pending.\n");
+    writeSlice(mDir, "S02", "# S02: Slice B\n\n**Goal:** B\n**Demo:** B\n\n## Tasks\n- [ ] **T01: Task** `est:10m`\n  Pending.\n");
+
+    const result = await runGSDDoctor(base, { fix: false });
+    assertTrue(
+      result.issues.some(i => i.code === "circular_slice_dependency"),
+      "detects circular dependency S01 → S02 → S01",
+    );
+    rmSync(base, { recursive: true, force: true });
+  }
+
+  // ── 2. Duplicate task IDs ──────────────────────────────────────────────────
+  console.log("\n=== duplicate task IDs ===");
+  {
+    const { base, mDir } = makeBase();
+    writeRoadmap(mDir, `# M001: Dup Test\n\n## Slices\n- [ ] **S01: Slice** \`risk:low\` \`depends:[]\`\n  > After this: done\n`);
+    writeSlice(mDir, "S01", "# S01: Slice\n\n**Goal:** G\n**Demo:** D\n\n## Tasks\n- [ ] **T01: First** `est:10m`\n  Task one.\n- [ ] **T01: Duplicate** `est:10m`\n  Task dup.\n");
+
+    const result = await runGSDDoctor(base, { fix: false });
+    assertTrue(
+      result.issues.some(i => i.code === "duplicate_task_id"),
+      "detects duplicate task ID T01",
+    );
+    rmSync(base, { recursive: true, force: true });
+  }
+
+  // ── 3. Orphaned slice directory ──────────────────────────────────────────
+  console.log("\n=== orphaned slice directory ===");
+  {
+    const { base, mDir } = makeBase();
+    writeRoadmap(mDir, `# M001: Orphan Test\n\n## Slices\n- [ ] **S01: Slice** \`risk:low\` \`depends:[]\`\n  > After this: done\n`);
+    writeSlice(mDir, "S01", "# S01: Slice\n\n**Goal:** G\n**Demo:** D\n\n## Tasks\n- [ ] **T01: Task** `est:10m`\n  Pending.\n");
+    // Create an extra slice directory not in roadmap
+    mkdirSync(join(mDir, "slices", "S99"), { recursive: true });
+
+    const result = await runGSDDoctor(base, { fix: false });
+    assertTrue(
+      result.issues.some(i => i.code === "orphaned_slice_directory" && i.message.includes("S99")),
+      "detects orphaned slice directory S99",
+    );
+    rmSync(base, { recursive: true, force: true });
+  }
+
+  // ── 4. Task file not in plan ───────────────────────────────────────────────
+  console.log("\n=== task file not in plan ===");
+  {
+    const { base, mDir } = makeBase();
+    writeRoadmap(mDir, `# M001: Extra Task Test\n\n## Slices\n- [ ] **S01: Slice** \`risk:low\` \`depends:[]\`\n  > After this: done\n`);
+    const sDir = writeSlice(mDir, "S01", "# S01: Slice\n\n**Goal:** G\n**Demo:** D\n\n## Tasks\n- [x] **T01: Task** `est:10m`\n  Done.\n");
+    // T01 summary (matches plan)
+    writeFileSync(join(sDir, "tasks", "T01-SUMMARY.md"), "---\nstatus: done\n---\n# T01\nDone.\n");
+    // T99 summary (NOT in plan)
+    writeFileSync(join(sDir, "tasks", "T99-SUMMARY.md"), "---\nstatus: done\n---\n# T99\nExtra.\n");
+
+    const result = await runGSDDoctor(base, { fix: false });
+    assertTrue(
+      result.issues.some(i => i.code === "task_file_not_in_plan" && i.message.includes("T99")),
+      "detects task summary T99 not in plan",
+    );
+    rmSync(base, { recursive: true, force: true });
+  }
+
+  // ── 5. Stale REPLAN file ────────────────────────────────────────────────────
+  console.log("\n=== stale REPLAN detection ===");
+  {
+    const { base, mDir } = makeBase();
+    writeRoadmap(mDir, `# M001: Replan Test\n\n## Slices\n- [ ] **S01: Slice** \`risk:low\` \`depends:[]\`\n  > After this: done\n`);
+    const sDir = writeSlice(mDir, "S01", "# S01: Slice\n\n**Goal:** G\n**Demo:** D\n\n## Tasks\n- [x] **T01: Task** `est:10m`\n  Done.\n");
+    writeFileSync(join(sDir, "tasks", "T01-SUMMARY.md"), "---\nstatus: done\ncompleted_at: 2026-01-01T00:00:00Z\n---\n# T01\nDone.\n");
+    // Add a REPLAN file even though all tasks are done
+    writeFileSync(join(sDir, "S01-REPLAN.md"), "# S01 REPLAN\nSomething changed.\n");
+
+    const result = await runGSDDoctor(base, { fix: false });
+    assertTrue(
+      result.issues.some(i => i.code === "stale_replan_file"),
+      "detects stale REPLAN when all tasks are done",
+    );
+    rmSync(base, { recursive: true, force: true });
+  }
+
+  // ── 6. Metrics ledger corrupt ───────────────────────────────────────────────
+  console.log("\n=== metrics ledger corrupt ===");
+  {
+    const { base, gsd, mDir } = makeBase();
+    writeRoadmap(mDir, `# M001: Metrics Test\n\n## Slices\n- [ ] **S01: Slice** \`risk:low\` \`depends:[]\`\n  > After this: done\n`);
+    writeSlice(mDir, "S01", "# S01: Slice\n\n**Goal:** G\n**Demo:** D\n\n## Tasks\n- [ ] **T01: Task** `est:10m`\n  Pending.\n");
+    // Write invalid metrics.json
+    writeFileSync(join(gsd, "metrics.json"), '{"version":2,"data":[]}');
+
+    const result = await runGSDDoctor(base, { fix: false });
+    assertTrue(
+      result.issues.some(i => i.code === "metrics_ledger_corrupt"),
+      "detects corrupt metrics ledger (version != 1)",
+    );
+    rmSync(base, { recursive: true, force: true });
+  }
+
+  // ── 7. Large planning file ──────────────────────────────────────────────────
+  console.log("\n=== large planning file ===");
+  {
+    const { base, mDir } = makeBase();
+    writeRoadmap(mDir, `# M001: Large File Test\n\n## Slices\n- [ ] **S01: Slice** \`risk:low\` \`depends:[]\`\n  > After this: done\n`);
+    const sDir = writeSlice(mDir, "S01", "# S01: Slice\n\n**Goal:** G\n**Demo:** D\n\n## Tasks\n- [ ] **T01: Task** `est:10m`\n  Pending.\n");
+    // Write a 101KB .md file
+    const bigContent = "# Big File\n" + "x".repeat(101 * 1024);
+    writeFileSync(join(sDir, "BIGFILE.md"), bigContent);
+
+    const result = await runGSDDoctor(base, { fix: false });
+    assertTrue(
+      result.issues.some(i => i.code === "large_planning_file"),
+      "detects large planning file over 100KB",
+    );
+    rmSync(base, { recursive: true, force: true });
+  }
+
+  // ── 8. Future timestamp ─────────────────────────────────────────────────────
+  console.log("\n=== future timestamp ===");
+  {
+    const { base, mDir } = makeBase();
+    writeRoadmap(mDir, `# M001: Timestamp Test\n\n## Slices\n- [ ] **S01: Slice** \`risk:low\` \`depends:[]\`\n  > After this: done\n`);
+    const sDir = writeSlice(mDir, "S01", "# S01: Slice\n\n**Goal:** G\n**Demo:** D\n\n## Tasks\n- [x] **T01: Task** `est:10m`\n  Done.\n");
+    // completed_at is 2 days in the future
+    const futureDate = new Date(Date.now() + 2 * 24 * 60 * 60 * 1000).toISOString();
+    writeFileSync(
+      join(sDir, "tasks", "T01-SUMMARY.md"),
+      `---\nstatus: done\ncompleted_at: ${futureDate}\n---\n# T01\nDone.\n`,
+    );
+
+    const result = await runGSDDoctor(base, { fix: false });
+    assertTrue(
+      result.issues.some(i => i.code === "future_timestamp"),
+      "detects future completed_at timestamp",
+    );
+    rmSync(base, { recursive: true, force: true });
+  }
+
+  // ── 9. JSON output format ───────────────────────────────────────────────────
+  console.log("\n=== JSON output format ===");
+  {
+    const { base, mDir } = makeBase();
+    writeRoadmap(mDir, `# M001: JSON Test\n\n## Slices\n- [ ] **S01: Slice** \`risk:low\` \`depends:[]\`\n  > After this: done\n`);
+    writeSlice(mDir, "S01", "# S01: Slice\n\n**Goal:** G\n**Demo:** D\n\n## Tasks\n- [ ] **T01: Task** `est:10m`\n  Pending.\n");
+
+    const result = await runGSDDoctor(base, { fix: false });
+    const json = formatDoctorReportJson(result);
+
+    let parsed: unknown;
+    try {
+      parsed = JSON.parse(json);
+    } catch {
+      parsed = null;
+    }
+
+    assertTrue(parsed !== null, "formatDoctorReportJson produces valid JSON");
+    assertTrue(typeof (parsed as Record<string, unknown>)?.ok === "boolean", "JSON has ok field");
+    assertTrue(Array.isArray((parsed as Record<string, unknown>)?.issues), "JSON has issues array");
+    assertTrue(Array.isArray((parsed as Record<string, unknown>)?.fixesApplied), "JSON has fixesApplied array");
+    assertTrue(typeof (parsed as Record<string, unknown>)?.generatedAt === "string", "JSON has generatedAt field");
+    assertTrue(typeof (parsed as Record<string, unknown>)?.summary === "object", "JSON has summary object");
+
+    rmSync(base, { recursive: true, force: true });
+  }
+
+  // ── 10. Dry-run mode ────────────────────────────────────────────────────────
+  console.log("\n=== dry-run mode ===");
+  {
+    const { base, mDir } = makeBase();
+    writeRoadmap(mDir, `# M001: Dry Run Test\n\n## Slices\n- [ ] **S01: Slice** \`risk:low\` \`depends:[]\`\n  > After this: done\n`);
+    const sDir = writeSlice(mDir, "S01", "# S01: Slice\n\n**Goal:** G\n**Demo:** D\n\n## Tasks\n- [x] **T01: Task** `est:10m`\n  Done.\n");
+
+    const result = await runGSDDoctor(base, { fix: true, dryRun: true });
+    // In dry-run mode, no actual files should be created
+    assertTrue(!existsSync(join(sDir, "S01-SUMMARY.md")), "dry-run does not create slice summary");
+    assertTrue(
+      result.fixesApplied.some(f => f.startsWith("[dry-run]")),
+      "dry-run mode reports would-fix entries",
+    );
+
+    rmSync(base, { recursive: true, force: true });
+  }
+
+  // ── 11. Per-check timing ─────────────────────────────────────────────────────
+  console.log("\n=== per-check timing ===");
+  {
+    const { base, mDir } = makeBase();
+    writeRoadmap(mDir, `# M001: Timing Test\n\n## Slices\n- [ ] **S01: Slice** \`risk:low\` \`depends:[]\`\n  > After this: done\n`);
+    writeSlice(mDir, "S01", "# S01: Slice\n\n**Goal:** G\n**Demo:** D\n\n## Tasks\n- [ ] **T01: Task** `est:10m`\n  Pending.\n");
+
+    const result = await runGSDDoctor(base, { fix: false });
+    assertTrue(result.timing !== undefined, "report includes timing");
+    assertTrue(typeof result.timing?.git === "number", "timing.git is a number");
+    assertTrue(typeof result.timing?.runtime === "number", "timing.runtime is a number");
+    assertTrue(typeof result.timing?.environment === "number", "timing.environment is a number");
+    assertTrue(typeof result.timing?.gsdState === "number", "timing.gsdState is a number");
+
+    rmSync(base, { recursive: true, force: true });
+  }
+
+  // ── 12. Doctor history ───────────────────────────────────────────────────────
+  console.log("\n=== doctor history ===");
+  {
+    const { base, gsd, mDir } = makeBase();
+    writeRoadmap(mDir, `# M001: History Test\n\n## Slices\n- [ ] **S01: Slice** \`risk:low\` \`depends:[]\`\n  > After this: done\n`);
+    writeSlice(mDir, "S01", "# S01: Slice\n\n**Goal:** G\n**Demo:** D\n\n## Tasks\n- [ ] **T01: Task** `est:10m`\n  Pending.\n");
+
+    await runGSDDoctor(base, { fix: false });
+
+    const historyPath = join(gsd, "doctor-history.jsonl");
+    assertTrue(existsSync(historyPath), "doctor-history.jsonl is created after run");
+
+    const { readDoctorHistory } = await import("../doctor.js");
+    const history = await readDoctorHistory(base);
+    assertTrue(history.length >= 1, "history has at least one entry");
+    assertTrue(typeof history[0]?.ts === "string", "history entry has ts field");
+    assertTrue(typeof history[0]?.ok === "boolean", "history entry has ok field");
+    assertTrue(typeof history[0]?.errors === "number", "history entry has errors count");
+    assertTrue(Array.isArray(history[0]?.codes), "history entry has codes array");
+
+    rmSync(base, { recursive: true, force: true });
+  }
+
+  report();
+}
+
+main().catch(err => {
+  console.error(err);
+  process.exit(1);
+});

package/src/resources/extensions/gsd/tests/doctor-providers.test.ts

@@ -47,6 +47,18 @@ function withEnv(vars: Record<string, string | undefined>, fn: () => void): void
   }
 }
 
+function withCwd(nextCwd: string, fn: () => void): void {
+  const saved = process.cwd();
+  process.chdir(nextCwd);
+  try {
+    fn();
+  } finally {
+    process.chdir(saved);
+  }
+}
+
+const PRESENT_TEST_VALUE = "configured";
+
 // ─── formatProviderReport ─────────────────────────────────────────────────────
 
 test("formatProviderReport returns fallback for empty results", () => {
@@ -312,7 +324,7 @@ test("runProviderChecks reports ok for Anthropic when GitHub Copilot env var is
   withEnv({
     ANTHROPIC_API_KEY: undefined,
     ANTHROPIC_OAUTH_TOKEN: undefined,
-    COPILOT_GITHUB_TOKEN: "ghu_copilot-token",
+    COPILOT_GITHUB_TOKEN: PRESENT_TEST_VALUE,
     GH_TOKEN: undefined,
     GITHUB_TOKEN: undefined,
     HOME: tmpHome,
@@ -336,7 +348,7 @@ test("runProviderChecks reports ok for Anthropic via GITHUB_TOKEN cross-provider
     ANTHROPIC_OAUTH_TOKEN: undefined,
     COPILOT_GITHUB_TOKEN: undefined,
     GH_TOKEN: undefined,
-    GITHUB_TOKEN: "ghp_github-token",
+    GITHUB_TOKEN: PRESENT_TEST_VALUE,
     HOME: tmpHome,
   }, () => {
     try {
@@ -354,7 +366,7 @@ test("runProviderChecks detects ANTHROPIC_OAUTH_TOKEN as valid Anthropic auth",
   const tmpHome = realpathSync(mkdtempSync(join(tmpdir(), "gsd-providers-oauth-test-")));
   withEnv({
     ANTHROPIC_API_KEY: undefined,
-    ANTHROPIC_OAUTH_TOKEN: "oauth-token-test",
+    ANTHROPIC_OAUTH_TOKEN: PRESENT_TEST_VALUE,
     COPILOT_GITHUB_TOKEN: undefined,
     GH_TOKEN: undefined,
     GITHUB_TOKEN: undefined,
@@ -401,3 +413,74 @@ test("runProviderChecks reports ok via Copilot auth.json for Anthropic", () => {
     rmSync(tmpHome, { recursive: true, force: true });
   });
 });
+
+test("runProviderChecks uses provider-qualified anthropic-vertex model IDs", () => {
+  const tmpHome = realpathSync(mkdtempSync(join(tmpdir(), "gsd-providers-vertex-prefix-home-")));
+  const repo = realpathSync(mkdtempSync(join(tmpdir(), "gsd-providers-vertex-prefix-repo-")));
+  mkdirSync(join(repo, ".gsd"), { recursive: true });
+  writeFileSync(
+    join(repo, ".gsd", "preferences.md"),
+    [
+      "---",
+      "models:",
+      "  execution: anthropic-vertex/claude-sonnet-4-6",
+      "---",
+      "",
+    ].join("\n"),
+  );
+
+  withEnv({
+    HOME: tmpHome,
+    ANTHROPIC_API_KEY: undefined,
+    ANTHROPIC_OAUTH_TOKEN: undefined,
+    ANTHROPIC_VERTEX_PROJECT_ID: "vertex-project",
+  }, () => {
+    withCwd(repo, () => {
+      const results = runProviderChecks();
+      const vertex = results.find(r => r.name === "anthropic-vertex");
+      const anthropic = results.find(r => r.name === "anthropic");
+      assert.ok(vertex, "anthropic-vertex result should exist");
+      assert.equal(vertex!.status, "ok", "should accept ANTHROPIC_VERTEX_PROJECT_ID as configured");
+      assert.ok(!anthropic || !anthropic.required, "plain anthropic should not be required for anthropic-vertex config");
+    });
+  });
+
+  rmSync(repo, { recursive: true, force: true });
+  rmSync(tmpHome, { recursive: true, force: true });
+});
+
+test("runProviderChecks uses object provider field for anthropic-vertex models", () => {
+  const tmpHome = realpathSync(mkdtempSync(join(tmpdir(), "gsd-providers-vertex-provider-home-")));
+  const repo = realpathSync(mkdtempSync(join(tmpdir(), "gsd-providers-vertex-provider-repo-")));
+  mkdirSync(join(repo, ".gsd"), { recursive: true });
+  writeFileSync(
+    join(repo, ".gsd", "preferences.md"),
+    [
+      "---",
+      "models:",
+      "  execution:",
+      "    model: claude-sonnet-4-6",
+      "    provider: anthropic-vertex",
+      "---",
+      "",
+    ].join("\n"),
+  );
+
+  withEnv({
+    HOME: tmpHome,
+    ANTHROPIC_API_KEY: undefined,
+    ANTHROPIC_OAUTH_TOKEN: undefined,
+    ANTHROPIC_VERTEX_PROJECT_ID: undefined,
+  }, () => {
+    withCwd(repo, () => {
+      const results = runProviderChecks();
+      const vertex = results.find(r => r.name === "anthropic-vertex");
+      assert.ok(vertex, "anthropic-vertex result should exist");
+      assert.equal(vertex!.status, "error", "missing vertex config should be reported against anthropic-vertex");
+      assert.ok(vertex!.detail?.includes("ANTHROPIC_VERTEX_PROJECT_ID"), "should point to vertex setup");
+    });
+  });
+
+  rmSync(repo, { recursive: true, force: true });
+  rmSync(tmpHome, { recursive: true, force: true });
+});

package/src/resources/extensions/gsd/tests/preferences.test.ts

@@ -208,30 +208,25 @@ test("git fields comprehensive validation", () => {
   assert.equal(preferences.git?.isolation, "branch");
 });
 
-test("auto_visualize, auto_report, compression_strategy, context_selection validate correctly", () => {
+test("auto_visualize, auto_report, context_selection validate correctly", () => {
   const { preferences, errors } = validatePreferences({
     auto_visualize: true,
     auto_report: false,
-    compression_strategy: "compress",
     context_selection: "smart",
   });
   assert.equal(errors.length, 0);
   assert.equal(preferences.auto_visualize, true);
   assert.equal(preferences.auto_report, false);
-  assert.equal(preferences.compression_strategy, "compress");
   assert.equal(preferences.context_selection, "smart");
 });
 
-test("auto_visualize, auto_report, compression_strategy, context_selection reject invalid values", () => {
+test("auto_visualize, auto_report, context_selection reject invalid values", () => {
   const { errors: e1 } = validatePreferences({ auto_visualize: "yes" as never });
   assert.ok(e1.some(e => e.includes("auto_visualize")));
 
   const { errors: e2 } = validatePreferences({ auto_report: 1 as never });
   assert.ok(e2.some(e => e.includes("auto_report")));
 
-  const { errors: e3 } = validatePreferences({ compression_strategy: "shrink" as never });
-  assert.ok(e3.some(e => e.includes("compression_strategy")));
-
   const { errors: e4 } = validatePreferences({ context_selection: "partial" as never });
   assert.ok(e4.some(e => e.includes("context_selection")));
 });

package/src/resources/extensions/gsd/tests/prompt-contracts.test.ts

@@ -0,0 +1,59 @@
+import test from "node:test";
+import assert from "node:assert/strict";
+import { readFileSync } from "node:fs";
+import { join } from "node:path";
+
+const promptsDir = join(process.cwd(), "src/resources/extensions/gsd/prompts");
+
+function readPrompt(name: string): string {
+  return readFileSync(join(promptsDir, `${name}.md`), "utf-8");
+}
+
+test("reactive-execute prompt keeps task summaries with subagents and avoids batch commits", () => {
+  const prompt = readPrompt("reactive-execute");
+  assert.match(prompt, /subagent-written summary as authoritative/i);
+  assert.match(prompt, /Do NOT create a batch commit/i);
+  assert.doesNotMatch(prompt, /\*\*Write task summaries\*\*/i);
+  assert.doesNotMatch(prompt, /\*\*Commit\*\* all changes/i);
+});
+
+test("run-uat prompt branches on dynamic UAT mode and supports runtime evidence", () => {
+  const prompt = readPrompt("run-uat");
+  assert.match(prompt, /\*\*Detected UAT mode:\*\*\s*`\{\{uatType\}\}`/);
+  assert.match(prompt, /uatType:\s*\{\{uatType\}\}/);
+  assert.match(prompt, /live-runtime/);
+  assert.match(prompt, /browser\/runtime\/network/i);
+  assert.match(prompt, /NEEDS-HUMAN/);
+  assert.doesNotMatch(prompt, /uatType:\s*artifact-driven/);
+});
+
+test("workflow-start prompt defaults to autonomy instead of per-phase confirmation", () => {
+  const prompt = readPrompt("workflow-start");
+  assert.match(prompt, /Keep moving by default/i);
+  assert.match(prompt, /Decision gates, not ceremony/i);
+  assert.doesNotMatch(prompt, /confirm with the user before proceeding/i);
+  assert.doesNotMatch(prompt, /Gate between phases/i);
+});
+
+test("discuss prompt allows implementation questions when they materially matter", () => {
+  const prompt = readPrompt("discuss");
+  assert.match(prompt, /Lead with experience, but ask implementation when it materially matters/i);
+  assert.match(prompt, /one gate, not two/i);
+  assert.doesNotMatch(prompt, /Questions must be about the experience, not the implementation/i);
+});
+
+test("guided discussion prompts avoid wrap-up prompts after every round", () => {
+  const milestonePrompt = readPrompt("guided-discuss-milestone");
+  const slicePrompt = readPrompt("guided-discuss-slice");
+  assert.match(milestonePrompt, /Do \*\*not\*\* ask a meta "ready to wrap up\?" question after every round/i);
+  assert.match(slicePrompt, /Do \*\*not\*\* ask a meta "ready to wrap up\?" question after every round/i);
+  assert.doesNotMatch(milestonePrompt, /I think I have a solid picture of this milestone\. Ready to wrap up/i);
+  assert.doesNotMatch(slicePrompt, /I think I have a solid picture of this slice\. Ready to wrap up/i);
+});
+
+test("guided-resume-task prompt preserves recovery state until work is superseded", () => {
+  const prompt = readPrompt("guided-resume-task");
+  assert.match(prompt, /Do \*\*not\*\* delete the continue file immediately/i);
+  assert.match(prompt, /successfully completed or you have written a newer summary\/continue artifact/i);
+  assert.doesNotMatch(prompt, /Delete the continue file after reading it/i);
+});

package/src/resources/extensions/gsd/tests/repo-identity-worktree.test.ts

@@ -3,7 +3,7 @@ import { join } from "node:path";
 import { tmpdir } from "node:os";
 import { execSync } from "node:child_process";
 
-import { externalGsdRoot, ensureGsdSymlink } from "../repo-identity.ts";
+import { repoIdentity, externalGsdRoot, ensureGsdSymlink, validateProjectId } from "../repo-identity.ts";
 import { createTestContext } from "./test-helpers.ts";
 
 const { assertEq, assertTrue, report } = createTestContext();
@@ -57,6 +57,26 @@ async function main(): Promise<void> {
     assertEq(preservedDirState, join(worktreePath, ".gsd"), "worktree .gsd directory is left in place for sync-based refresh");
     assertTrue(lstatSync(join(worktreePath, ".gsd")).isDirectory(), "worktree .gsd directory remains a directory");
     assertTrue(existsSync(join(worktreePath, ".gsd", "milestones", "stale.txt")), "existing worktree .gsd directory contents remain available for sync logic");
+
+    console.log("\n=== GSD_PROJECT_ID overrides computed repo hash ===");
+    process.env.GSD_PROJECT_ID = "my-project";
+    assertEq(repoIdentity(base), "my-project", "repoIdentity returns GSD_PROJECT_ID when set");
+    assertEq(externalGsdRoot(base), join(stateDir, "projects", "my-project"), "externalGsdRoot uses GSD_PROJECT_ID");
+    delete process.env.GSD_PROJECT_ID;
+
+    console.log("\n=== GSD_PROJECT_ID falls back to hash when unset ===");
+    const hashIdentity = repoIdentity(base);
+    assertTrue(/^[0-9a-f]{12}$/.test(hashIdentity), "repoIdentity returns 12-char hex hash when GSD_PROJECT_ID is unset");
+
+    console.log("\n=== validateProjectId rejects invalid values ===");
+    for (const invalid of ["has spaces", "path/traversal", "dot..dot", "back\\slash"]) {
+      assertTrue(!validateProjectId(invalid), `validateProjectId rejects invalid value: "${invalid}"`);
+    }
+
+    console.log("\n=== validateProjectId accepts valid values ===");
+    for (const valid of ["my-project", "foo_bar", "abc123", "A-Z_0-9"]) {
+      assertTrue(validateProjectId(valid), `validateProjectId accepts valid value: "${valid}"`);
+    }
   } finally {
     delete process.env.GSD_STATE_DIR;
     rmSync(base, { recursive: true, force: true });

package/src/resources/extensions/gsd/tests/run-uat.test.ts

@@ -210,7 +210,7 @@ async function main(): Promise<void> {
   const sliceId = 'S01';
   const uatPath = '.gsd/milestones/M001/slices/S01/S01-UAT.md';
   const uatResultPath = '.gsd/milestones/M001/slices/S01/S01-UAT-RESULT.md';
-  const uatType = 'artifact-driven';
+  const uatType = 'live-runtime';
   const inlinedContext = '<!-- no context -->';
 
   let promptResult: string | undefined;
@@ -246,13 +246,21 @@ async function main(): Promise<void> {
     promptResult?.includes(uatResultPath) ?? false,
     `prompt contains uatResultPath value after substitution`,
   );
+  assertTrue(
+    promptResult?.includes(`Detected UAT mode:** \`${uatType}\``) ?? false,
+    `prompt contains detected dynamic uatType value "${uatType}" after substitution`,
+  );
+  assertTrue(
+    promptResult?.includes(`uatType: ${uatType}`) ?? false,
+    `prompt contains dynamic uatType frontmatter value "${uatType}" after substitution`,
+  );
   assertTrue(
     !/\{\{[^}]+\}\}/.test(promptResult ?? ''),
     'no unreplaced {{...}} tokens remain after variable substitution',
   );
   assertTrue(
-    /artifact|execute|run/i.test(promptResult ?? ''),
-    'prompt contains artifact-driven execution language (artifact/execute/run)',
+    /browser|runtime|execute|run/i.test(promptResult ?? ''),
+    'prompt contains runtime execution language (browser/runtime/execute/run)',
   );
   assertTrue(
     !/surfaced for human review/i.test(promptResult ?? ''),