@opengsd/gsd-pi 1.2.0-dev.844675c9 → 1.2.0-dev.955e4da0

package/src/resources/extensions/gsd/safety/file-change-validator.ts

@@ -39,6 +39,20 @@ export interface FileChangeAudit {
 
 // ─── Public API ─────────────────────────────────────────────────────────────
 
+/**
+ * Build the effective allowlist for a unit's file-change audit.
+ *
+ * When GSD manages .gitignore (manage_gitignore unset or true), ensureGitignore()
+ * appends baseline patterns at auto-start and the edit rides into the task's
+ * auto-commit — so .gitignore changes must not be attributed to the task.
+ */
+export function effectiveFileChangeAllowlist(
+  baseAllowlist: string[],
+  manageGitignore: boolean | undefined,
+): string[] {
+  return manageGitignore === false ? baseAllowlist : [...baseAllowlist, ".gitignore"];
+}
+
 /**
  * Validate file changes after auto-commit for an execute-task unit.
  * Returns null if task data is unavailable or DB is not loaded.

package/src/resources/extensions/gsd/slice-parallel-orchestrator.ts

@@ -30,7 +30,8 @@ import { isAbsolute, join, dirname, resolve } from "node:path";
 import { fileURLToPath } from "node:url";
 import { gsdRoot } from "./paths.js";
 import { createWorktree, worktreePath, removeWorktree } from "./worktree-manager.js";
-import { autoWorktreeBranch, runWorktreePostCreateHook, syncGsdStateToWorktree } from "./auto-worktree.js";
+import { autoWorktreeBranch, runWorktreePostCreateHook, syncGsdStateToWorktreeByScope } from "./auto-worktree.js";
+import { createWorkspace, scopeMilestone } from "./workspace.js";
 import {
   writeSessionStatus,
   removeSessionStatus,
@@ -163,7 +164,10 @@ function createSliceWorktree(basePath: string, milestoneId: string, sliceId: str
   if (hookError) {
     throw new Error(`slice worktree post-create hook failed (${wtName}): ${hookError}`);
   }
-  syncGsdStateToWorktree(basePath, wtPath);
+  syncGsdStateToWorktreeByScope(
+    scopeMilestone(createWorkspace(basePath), milestoneId),
+    scopeMilestone(createWorkspace(wtPath), milestoneId),
+  );
 
   if (!existsSync(join(wtPath, ".gsd"))) {
     throw new Error(`slice worktree preflight failed (${wtName}): missing .gsd in worktree`);

package/src/resources/extensions/gsd/state-transition-matrix.ts

@@ -131,6 +131,48 @@ export function findTransition(
   );
 }
 
+/**
+ * Edge-keyed legality check for the Phase Transition Invariant (ADR-030).
+ * `advance()` derives the next Phase and asserts the (from → to) edge here.
+ *
+ * The matrix is an assertion, not a decision-maker — `deriveState` already
+ * chose the Phase. A self-edge is trivially legal (no transition to assert). An
+ * edge is legal when some matrix entry permits it, honoring the `*` wildcard
+ * rows (e.g. any → blocked via manual-block, any → executing via
+ * retryable-failure).
+ *
+ * Note: the matrix is currently a sparse hardening spec, not a complete
+ * legal-edge graph, so `advance()` consumes this in advisory mode (telemetry
+ * only). It must be expanded to cover every edge `deriveState` emits before
+ * enforcement flips on.
+ */
+export function isLegalEdge(from: Phase, to: Phase): boolean {
+  if (from === to) return true;
+  return STATE_TRANSITION_MATRIX.some(
+    (entry) => (entry.from === from || entry.from === "*") && entry.to === to,
+  );
+}
+
+/**
+ * Thrown when an illegal derived Phase edge survives reconciliation. Carries
+ * both endpoints so Recovery Classification can report them. Recognized by
+ * class in `classifyFailure` and mapped to the `illegal-transition` kind.
+ */
+export class IllegalPhaseTransitionError extends Error {
+  // Explicit fields, not constructor parameter properties — strip-types
+  // consumers (workspace-index subprocess, integration tests) reject the
+  // parameter-property syntax with ERR_UNSUPPORTED_TYPESCRIPT_SYNTAX.
+  readonly from: Phase;
+  readonly to: Phase;
+
+  constructor(from: Phase, to: Phase) {
+    super(`Illegal phase transition ${from} -> ${to} survived reconciliation`);
+    this.name = "IllegalPhaseTransitionError";
+    this.from = from;
+    this.to = to;
+  }
+}
+
 export function validateTransitionMatrix(requiredEvents: readonly string[]): MatrixValidationResult {
   const seen = new Set<string>();
   const duplicateKeys: string[] = [];

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

@@ -52,7 +52,6 @@ import { isTerminalMilestoneSummaryContent } from './milestone-summary-classifie
 
 import {
   isDbAvailable,
-  wasDbOpenAttempted,
   getAllMilestones,
   getMilestone,
   getMilestoneSlices,
@@ -63,9 +62,14 @@ import {
   getLatestAssessmentByScope,
   getPendingGateCountForTurn,
 } from './gsd-db.js';
+import { wasWorkflowDatabaseOpenAttempted } from './db-workspace.js';
 import { formatCompletePhaseNextAction, countUnmappedActiveRequirements } from './requirements-backlog.js';
 import type { MilestoneRow } from './db-milestone-artifact-rows.js';
 import type { SliceRow, TaskRow } from './db-task-slice-rows.js';
+import {
+  classifyMilestoneReadiness,
+  readinessNeedsDiscussion,
+} from './milestone-readiness.js';
 
 function formatNeedsAttentionBlocker(milestoneId: string): string {
   return [
@@ -395,7 +399,7 @@ export async function deriveState(
     stopDbTimer({ phase: result.phase, milestone: result.activeMilestone?.id });
     _telemetry.dbDeriveCount++;
   } else {
-    if (wasDbOpenAttempted()) {
+    if (wasWorkflowDatabaseOpenAttempted()) {
       logWarning("state", "DB unavailable — refusing implicit markdown state derivation");
     }
     result = {
@@ -507,6 +511,12 @@ async function buildRegistryAndFindActive(
     const title = stripMilestonePrefix(m.title) || m.id;
     const hasContext = !!resolveMilestoneFile(basePath, m.id, "CONTEXT");
     const hasDraftContext = !hasContext && !!resolveMilestoneFile(basePath, m.id, "CONTEXT-DRAFT");
+    const readiness = classifyMilestoneReadiness({
+      status: m.status,
+      hasContext,
+      hasDraftContext,
+      sliceCount: slices.length,
+    });
 
     if (!activeMilestoneFound) {
       const deps = m.depends_on;
@@ -517,9 +527,9 @@ async function buildRegistryAndFindActive(
         continue;
       }
 
-      if (m.status === 'queued' && slices.length === 0 && !hasContext) {
+      if (readiness.kind === 'queued-shell') {
         if (!firstDeferredQueuedShell) {
-          firstDeferredQueuedShell = { id: m.id, title, deps, hasDraftContext };
+          firstDeferredQueuedShell = { id: m.id, title, deps, hasDraftContext: readiness.hasDraftContext };
         }
         registry.push({ id: m.id, title, status: 'pending', ...(deps.length > 0 ? { dependsOn: deps } : {}) });
         continue;
@@ -533,7 +543,7 @@ async function buildRegistryAndFindActive(
         continue;
       }
 
-      if ((m.status === 'needs-discussion' && !hasContext) || hasDraftContext) activeMilestoneHasDraft = true;
+      if (readinessNeedsDiscussion(readiness)) activeMilestoneHasDraft = true;
 
       activeMilestone = { id: m.id, title };
       activeMilestoneSlices = slices;

package/src/resources/extensions/gsd/status-guards.ts

@@ -1,17 +1,68 @@
 /**
- * Status predicates for GSD state-machine guards.
+ * Status predicates and the canonical status vocabulary for GSD state-machine
+ * guards (ADR-030).
  *
- * The DB stores status as free-form strings. Three values indicate
- * "closed": "complete" (canonical), "done" (legacy / alias),
- * "closed" (legacy/imported), and
- * "skipped" (user-directed skip via rethink or backtrack).
- * Every inline `status === "complete" || status === "done"` should
- * use isClosedStatus() instead.
+ * The DB column is free-form `string` so legacy/imported rows still load. Three
+ * raw values besides canonical "complete"/"skipped" indicate "closed": "done"
+ * (legacy alias), "closed" (legacy/imported), and "skipped" (user-directed skip
+ * via rethink or backtrack). `RAW_CLOSED_STATUSES` is the single source for both
+ * `isClosedStatus()` and the SQL terminal-status fragment
+ * (`db/sql-constants.ts` derives `TERMINAL_STATUS_SQL` from it), replacing the
+ * prior independent definitions.
+ *
+ * `toStatus()` is the single seam where a free-form string becomes the canonical
+ * `Status` vocabulary; the Status Transition Core writes canonical, so the store
+ * converges over time without a forced migration.
+ */
+
+/**
+ * Canonical, normalized entity-status vocabulary across milestones, slices, and
+ * tasks — the single source for both the `Status` type and the runtime
+ * membership set. The in-memory domain speaks `Status`; the DB column stays
+ * free-form.
+ */
+export const CANONICAL_STATUSES = [
+  "pending", "queued", "active", "parked", "in_progress", "blocked", "complete", "skipped", "deferred",
+] as const;
+export type Status = (typeof CANONICAL_STATUSES)[number];
+const CANONICAL_STATUS_SET: ReadonlySet<string> = new Set(CANONICAL_STATUSES);
+
+/**
+ * Raw status values that mean a unit is closed — the single source of truth.
+ * Includes legacy/imported aliases ("done", "closed") alongside canonical
+ * "complete"/"skipped" because the DB column is free-form and older rows /
+ * imports still carry them. Order matters: `TERMINAL_STATUS_SQL` is derived
+ * from this array verbatim.
+ */
+export const RAW_CLOSED_STATUSES = ["complete", "done", "skipped", "closed"] as const;
+const RAW_CLOSED_SET: ReadonlySet<string> = new Set(RAW_CLOSED_STATUSES);
+
+/** Free-form aliases mapped to their canonical Status on read. */
+const ALIAS_TO_CANONICAL: Readonly<Record<string, Status>> = {
+  done: "complete",
+  closed: "complete",
+  planned: "pending",
+  "in-progress": "in_progress",
+};
+
+/**
+ * Normalize a free-form DB status string into the canonical `Status`
+ * vocabulary. Maps known aliases (done/closed → complete, planned → pending,
+ * in-progress → in_progress). An unrecognized/legacy value is **quarantined** —
+ * preserved verbatim rather than silently remapped to a wrong canonical state —
+ * so reads never fail and reconciliation/telemetry can surface it.
  */
+export function toStatus(raw: string): Status {
+  const value = raw.trim();
+  if (CANONICAL_STATUS_SET.has(value)) return value as Status;
+  const alias = ALIAS_TO_CANONICAL[value];
+  if (alias) return alias;
+  return value as Status;
+}
 
 /** Returns true when a milestone, slice, or task status indicates closure. */
 export function isClosedStatus(status: string): boolean {
-  return status === "complete" || status === "done" || status === "skipped" || status === "closed";
+  return RAW_CLOSED_SET.has(status);
 }
 
 /** Returns true when a slice status indicates it was deferred by a decision. */

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

@@ -46,6 +46,13 @@
      For simple projects or scripts:
      - Executable verification commands (bash assertions, curl checks, etc.) are sufficient
 
+     For browser-facing web apps (React, Next, Vue, Vite, static HTML, etc.):
+     - Prefer `npx playwright test <spec>` in Verification when `@playwright/test` exists
+     - If Playwright is not set up yet, the first UI slice should add `playwright.config.ts`,
+       a minimal smoke spec (for example `e2e/smoke.spec.ts`), and a safe verify command
+     - Slice closeout UAT must declare `browser-executable` or `runtime-executable` — not
+       `artifact-driven` — when checks navigate to localhost or exercise the live UI
+
      If the project has no test framework and the work is non-trivial,
      the first task should set one up. A test runner costs 2 minutes
      and pays for itself immediately.

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

@@ -20,6 +20,7 @@
 
 - **Complexity:** {{simple | complex}}
 - **Why:** {{one-line rationale citing the signals that decided it}}
+- **Web stack:** {{browser-facing framework or "not a web UI" — e.g. Next.js + Playwright, static HTML, N/A for CLI/API-only}}
 
 ## Current State
 

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

@@ -42,7 +42,7 @@
 - Contract verification: {{tests / shell verifiers / fixtures / artifact checks}}
 - Integration verification: {{real subsystem interaction that must be exercised, or none}}
 - Operational verification: {{service lifecycle / restart / reconnect / supervision / deploy-install behavior, or none}}
-- UAT / human verification: {{what needs real human judgment, or none}}
+- UAT / human verification: {{what needs real human judgment, or none — for web apps prefer browser-executable / runtime-executable Playwright checks over deferring UI proof to humans}}
 
 ## Milestone Definition of Done
 

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

@@ -5,9 +5,13 @@
 
 ## UAT Type
 
-- UAT mode: {{artifact-driven | live-runtime | human-experience | mixed}}
+- UAT mode: {{artifact-driven | browser-executable | runtime-executable | live-runtime | human-experience | mixed}}
 - Why this mode is sufficient: {{reason}}
 
+<!-- Web apps: never use artifact-driven when steps open a browser, navigate to localhost,
+     or capture screenshots. Use browser-executable (browser_* tools) or runtime-executable
+     (npx playwright test). Name dev-server preconditions below. -->
+
 ## Preconditions
 
 {{whatMustBeTrueBeforeTesting — server running, data seeded, etc.}}

package/src/resources/extensions/gsd/tests/ask-user-questions-render.test.ts

@@ -0,0 +1,92 @@
+/**
+ * Regression tests for ask_user_questions renderResult when Claude Code MCP
+ * omits structuredContent and the chat controller passes details as undefined
+ * or an empty object.
+ */
+
+import assert from "node:assert/strict";
+import { test } from "node:test";
+import askUserQuestionsExtension from "../../ask-user-questions.ts";
+
+const fakeTheme = {
+	fg: (_color: string, text: string) => text,
+	bold: (text: string) => text,
+};
+
+function getAskUserQuestionsTool() {
+	const tools: any[] = [];
+	askUserQuestionsExtension({ registerTool: (tool: any) => tools.push(tool) } as any);
+	const tool = tools.find((t) => t.name === "ask_user_questions");
+	assert.ok(tool, "ask_user_questions should be registered");
+	return tool;
+}
+
+function renderText(tool: any, result: unknown, args?: unknown): string {
+	const rendered = tool.renderResult(result, {}, fakeTheme, { args });
+	return String((rendered as any).content ?? (rendered as any).text ?? rendered);
+}
+
+const depthQuestion = {
+	id: "depth_check",
+	header: "Depth Check",
+	question: "Did I capture the depth right?",
+	options: [
+		{ label: "Yes, you got it", description: "Proceed" },
+		{ label: "Not quite - let me clarify", description: "Clarify" },
+	],
+};
+
+test("ask_user_questions renderResult shows answers when details is undefined (MCP text-only wire)", () => {
+	const tool = getAskUserQuestionsTool();
+	const contentText = JSON.stringify({
+		answers: { depth_check: { answers: ["Yes, you got it"] } },
+	});
+	const text = renderText(
+		tool,
+		{
+			content: [{ type: "text", text: contentText }],
+			details: undefined,
+			isError: false,
+		},
+		{ questions: [depthQuestion] },
+	);
+
+	assert.match(text, /✓/);
+	assert.match(text, /Depth Check/);
+	assert.match(text, /Yes, you got it/);
+	assert.doesNotMatch(text, /Cancelled/i);
+});
+
+test("ask_user_questions renderResult shows answers when details is empty object (#cc-elicitation)", () => {
+	const tool = getAskUserQuestionsTool();
+	const contentText = JSON.stringify({
+		answers: { depth_check: { answers: ["Yes, you got it"] } },
+	});
+	const text = renderText(
+		tool,
+		{
+			content: [{ type: "text", text: contentText }],
+			details: {},
+			isError: false,
+		},
+		{ questions: [depthQuestion] },
+	);
+
+	assert.match(text, /Yes, you got it/);
+	assert.doesNotMatch(text, /Cancelled/i);
+});
+
+test("ask_user_questions renderResult still shows Cancelled for explicit cancel payload", () => {
+	const tool = getAskUserQuestionsTool();
+	const text = renderText(
+		tool,
+		{
+			content: [{ type: "text", text: "ask_user_questions was cancelled before receiving a response" }],
+			details: { questions: [depthQuestion], response: null, cancelled: true },
+			isError: false,
+		},
+		{ questions: [depthQuestion] },
+	);
+
+	assert.match(text, /Cancelled/i);
+});

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

@@ -29,11 +29,16 @@ import {
   _refreshLastCommitForTests,
   _getLastCommitForTests,
   _getLastCommitFetchedAtForTests,
+  formatToolSurfaceSnapshot,
   formatRuntimeHealthSignal,
   shouldRenderRoadmapProgress,
 } from "../auto-dashboard.ts";
 import { getAutoDashboardData } from "../auto.ts";
-import { autoSession } from "../auto-runtime-state.ts";
+import {
+  autoSession,
+  clearAutoToolSurfaceSnapshot,
+  recordAutoToolSurfaceSnapshot,
+} from "../auto-runtime-state.ts";
 import { formatRtkSavingsLabel } from "../../shared/rtk-session-stats.ts";
 import {
   openDatabase,
@@ -530,6 +535,29 @@ test("getAutoDashboardData returns RTK savings in the dashboard payload", () =>
   }
 });
 
+test("getAutoDashboardData exposes typed tool-surface snapshots", () => {
+  autoSession.reset();
+  clearAutoToolSurfaceSnapshot();
+  autoSession.active = true;
+  recordAutoToolSurfaceSnapshot({
+    source: "provider-adjustment",
+    unitType: "run-uat",
+    modelFacingToolNames: ["read"],
+    registeredToolNames: ["read", "browser_navigate"],
+    scopedToolNames: ["read", "browser_navigate"],
+    presentedToolNames: ["browser_navigate"],
+    capturedAt: 456,
+  });
+
+  const data = getAutoDashboardData();
+
+  assert.equal(data.toolSurface?.source, "provider-adjustment");
+  assert.equal(formatToolSurfaceSnapshot(data.toolSurface), "run-uat: model 1 / registered 2 / scoped 2 / presented 1");
+
+  autoSession.reset();
+  clearAutoToolSurfaceSnapshot();
+});
+
 test("RTK savings label formats the dashboard footer text", () => {
   assert.equal(formatRtkSavingsLabel(null), null);
   assert.equal(