gsd-pi 2.57.0-dev.f22a903 → 2.58.0-dev.778d6ac

package/dist/resources/extensions/gsd/model-router.js

@@ -11,6 +11,12 @@ const MODEL_CAPABILITY_TIER = {
     "claude-3-5-haiku-latest": "light",
     "claude-3-haiku-20240307": "light",
     "gpt-4o-mini": "light",
+    "gpt-4.1-mini": "light",
+    "gpt-4.1-nano": "light",
+    "gpt-5-mini": "light",
+    "gpt-5-nano": "light",
+    "gpt-5.1-codex-mini": "light",
+    "gpt-5.3-codex-spark": "light",
     "gemini-2.0-flash": "light",
     "gemini-flash-2.0": "light",
     // Standard-tier models
@@ -18,14 +24,25 @@ const MODEL_CAPABILITY_TIER = {
     "claude-sonnet-4-5-20250514": "standard",
     "claude-3-5-sonnet-latest": "standard",
     "gpt-4o": "standard",
+    "gpt-4.1": "standard",
+    "gpt-5.1-codex-max": "standard",
     "gemini-2.5-pro": "standard",
     "deepseek-chat": "standard",
     // Heavy-tier models (most capable)
     "claude-opus-4-6": "heavy",
     "claude-3-opus-latest": "heavy",
     "gpt-4-turbo": "heavy",
+    "gpt-5": "heavy",
+    "gpt-5-pro": "heavy",
+    "gpt-5.1": "heavy",
+    "gpt-5.2": "heavy",
+    "gpt-5.2-codex": "heavy",
+    "gpt-5.3-codex": "heavy",
+    "gpt-5.4": "heavy",
     "o1": "heavy",
     "o3": "heavy",
+    "o4-mini": "heavy",
+    "o4-mini-deep-research": "heavy",
 };
 // ─── Cost Table (per 1K input tokens, approximate USD) ───────────────────────
 // Used for cross-provider cost comparison when multiple providers offer
@@ -38,6 +55,23 @@ const MODEL_COST_PER_1K_INPUT = {
     "claude-opus-4-6": 0.015,
     "gpt-4o-mini": 0.00015,
     "gpt-4o": 0.0025,
+    "gpt-4.1": 0.002,
+    "gpt-4.1-mini": 0.0004,
+    "gpt-4.1-nano": 0.0001,
+    "gpt-5": 0.01,
+    "gpt-5-mini": 0.0003,
+    "gpt-5-nano": 0.0001,
+    "gpt-5-pro": 0.015,
+    "gpt-5.1": 0.005,
+    "gpt-5.1-codex-max": 0.003,
+    "gpt-5.1-codex-mini": 0.0003,
+    "gpt-5.2": 0.005,
+    "gpt-5.2-codex": 0.005,
+    "gpt-5.3-codex": 0.005,
+    "gpt-5.3-codex-spark": 0.0003,
+    "gpt-5.4": 0.005,
+    "o4-mini": 0.005,
+    "o4-mini-deep-research": 0.005,
     "gemini-2.0-flash": 0.0001,
     "gemini-2.5-pro": 0.00125,
     "deepseek-chat": 0.00014,
@@ -133,7 +167,7 @@ export function escalateTier(currentTier) {
  */
 export function defaultRoutingConfig() {
     return {
-        enabled: false,
+        enabled: true,
         escalate_on_failure: true,
         budget_pressure: true,
         cross_provider: true,

package/dist/resources/extensions/gsd/native-git-bridge.js

@@ -758,6 +758,23 @@ export function nativeResetHard(basePath) {
     }
     execSync("git reset --hard HEAD", { cwd: basePath, stdio: "pipe" });
 }
+/**
+ * Get the subject line of a commit (git log -1 --format=%s <ref>).
+ * Returns empty string if the ref doesn't exist.
+ */
+export function nativeCommitSubject(basePath, ref) {
+    try {
+        return execFileSync("git", ["log", "-1", "--format=%s", ref], {
+            cwd: basePath,
+            stdio: ["ignore", "pipe", "pipe"],
+            encoding: "utf-8",
+            env: GIT_NO_PROMPT_ENV,
+        }).trim();
+    }
+    catch {
+        return "";
+    }
+}
 /**
  * Delete a branch.
  * Native: libgit2 branch delete.

package/dist/resources/extensions/gsd/notifications.js

@@ -7,7 +7,12 @@ import { CmuxClient, emitOsc777Notification, resolveCmuxConfig } from "../cmux/i
  * Send a native desktop notification. Non-blocking, non-fatal.
  * macOS: osascript, Linux: notify-send, Windows: skipped.
  */
-export function sendDesktopNotification(title, message, level = "info", kind = "complete") {
+export function sendDesktopNotification(title, message, level = "info", kind = "complete", projectName) {
+    // When a projectName is provided and the title is the default "GSD",
+    // replace it with a project-qualified title for multi-project clarity.
+    if (projectName && title === "GSD") {
+        title = formatNotificationTitle(projectName);
+    }
     const loaded = loadEffectiveGSDPreferences()?.preferences;
     if (!shouldSendDesktopNotification(kind, loaded?.notifications))
         return;
@@ -45,6 +50,16 @@ export function shouldSendDesktopNotification(kind, preferences = loadEffectiveG
             return preferences?.on_complete ?? true;
     }
 }
+/**
+ * Format a notification title that includes the project name for context.
+ * Returns "GSD — projectName" when a project name is available, otherwise "GSD".
+ */
+export function formatNotificationTitle(projectName) {
+    const trimmed = projectName?.trim();
+    if (trimmed)
+        return `GSD — ${trimmed}`;
+    return "GSD";
+}
 export function buildDesktopNotificationCommand(platform, title, message, level = "info") {
     const normalizedTitle = normalizeNotificationText(title);
     const normalizedMessage = normalizeNotificationText(message);

package/dist/resources/extensions/gsd/parallel-eligibility.js

@@ -74,7 +74,18 @@ export async function analyzeParallelEligibility(basePath) {
     for (const mid of milestoneIds) {
         const entry = registryMap.get(mid);
         const title = entry?.title ?? mid;
-        const status = entry?.status ?? "pending";
+        // Rule 0: milestones with no registry entry (ghost directories, unknown
+        // state) are ineligible — we cannot determine their status or deps (#2501)
+        if (!entry) {
+            ineligible.push({
+                milestoneId: mid,
+                title,
+                eligible: false,
+                reason: "Milestone has no planning data — cannot determine eligibility.",
+            });
+            continue;
+        }
+        const status = entry.status;
         // Rule 1: skip complete and parked milestones
         if (status === "complete" || status === "parked") {
             ineligible.push({
@@ -86,7 +97,7 @@ export async function analyzeParallelEligibility(basePath) {
             continue;
         }
         // Rule 2: check dependency satisfaction
-        const deps = entry?.dependsOn ?? [];
+        const deps = entry.dependsOn ?? [];
         const unsatisfied = deps.filter(dep => {
             const depEntry = registryMap.get(dep);
             return !depEntry || depEntry.status !== "complete";

package/dist/resources/extensions/gsd/parallel-merge.js

@@ -4,6 +4,9 @@
  * Handles merging completed milestone worktrees back to main branch
  * with safety checks for parallel execution context.
  */
+import { existsSync, readdirSync } from "node:fs";
+import { join } from "node:path";
+import { spawnSync } from "node:child_process";
 import { loadFile } from "./files.js";
 import { resolveMilestoneFile } from "./paths.js";
 import { mergeMilestoneToMain } from "./auto-worktree.js";
@@ -11,19 +14,89 @@ import { MergeConflictError } from "./git-service.js";
 import { removeSessionStatus } from "./session-status-io.js";
 import { getErrorMessage } from "./error-utils.js";
 // ─── Merge Queue ───────────────────────────────────────────────────────────
+/**
+ * Check whether a milestone is complete by querying its worktree SQLite DB.
+ * Uses a subprocess to avoid disrupting the global DB singleton.
+ * Returns true when milestones.status = 'complete' in the worktree's gsd.db.
+ */
+export function isMilestoneCompleteInWorktreeDb(basePath, mid) {
+    const dbPath = join(basePath, ".gsd", "worktrees", mid, ".gsd", "gsd.db");
+    if (!existsSync(dbPath))
+        return false;
+    try {
+        const result = spawnSync("sqlite3", [dbPath, `SELECT status FROM milestones WHERE id='${mid}' LIMIT 1`], { timeout: 3000, encoding: "utf-8" });
+        return (result.stdout || "").trim() === "complete";
+    }
+    catch {
+        return false;
+    }
+}
+/**
+ * Discover milestone IDs with status='complete' in their worktree DB,
+ * scanning .gsd/worktrees/<MID>/.gsd/gsd.db for each worktree directory.
+ */
+function discoverDbCompletedMilestones(basePath) {
+    const completed = new Set();
+    const worktreeDir = join(basePath, ".gsd", "worktrees");
+    try {
+        for (const entry of readdirSync(worktreeDir)) {
+            if (entry.startsWith("M") && isMilestoneCompleteInWorktreeDb(basePath, entry)) {
+                completed.add(entry);
+            }
+        }
+    }
+    catch {
+        // worktrees dir may not exist
+    }
+    return completed;
+}
 /**
  * Determine safe merge order for completed milestones.
  * Sequential: merge in milestone ID order (M001 before M002).
  * By-completion: merge in the order milestones finished.
+ *
+ * When basePath is provided, also checks worktree SQLite DBs as the
+ * source of truth — workers with stale orchestrator state (e.g. "error")
+ * are included if their worktree DB shows status='complete'.
+ * See: https://github.com/gsd-build/gsd-2/issues/2812
  */
-export function determineMergeOrder(workers, order = "sequential") {
-    const completed = workers.filter(w => w.state === "stopped");
+export function determineMergeOrder(workers, order = "sequential", basePath) {
+    // Start with workers the orchestrator already knows are stopped
+    const stoppedIds = new Set(workers.filter(w => w.state === "stopped").map(w => w.milestoneId));
+    // When basePath is available, also check worktree DBs for milestones
+    // whose orchestrator state is stale but are actually complete (#2812)
+    const dbCompleted = basePath ? discoverDbCompletedMilestones(basePath) : new Set();
+    // Union: milestone is mergeable if stopped OR DB-complete
+    const mergeableIds = new Set([...stoppedIds, ...dbCompleted]);
+    // Build the list from tracked workers + any DB-discovered milestones
+    // not tracked by the orchestrator at all
+    const workerMap = new Map(workers.map(w => [w.milestoneId, w]));
+    const allMergeable = [];
+    for (const mid of mergeableIds) {
+        const w = workerMap.get(mid);
+        if (w) {
+            allMergeable.push(w);
+        }
+        else {
+            // Milestone discovered from worktree DB but not in workers list
+            allMergeable.push({
+                milestoneId: mid,
+                title: mid,
+                pid: 0,
+                process: null,
+                worktreePath: basePath ? join(basePath, ".gsd", "worktrees", mid) : "",
+                startedAt: 0,
+                state: "stopped",
+                cost: 0,
+            });
+        }
+    }
     if (order === "by-completion") {
-        return completed
+        return allMergeable
             .sort((a, b) => a.startedAt - b.startedAt) // earliest first
             .map(w => w.milestoneId);
     }
-    return completed
+    return allMergeable
         .sort((a, b) => a.milestoneId.localeCompare(b.milestoneId))
         .map(w => w.milestoneId);
 }
@@ -82,7 +155,7 @@ export async function mergeCompletedMilestone(basePath, milestoneId) {
  * Stops on first conflict and returns results so far.
  */
 export async function mergeAllCompleted(basePath, workers, order = "sequential") {
-    const mergeOrder = determineMergeOrder(workers, order);
+    const mergeOrder = determineMergeOrder(workers, order, basePath);
     const results = [];
     for (const mid of mergeOrder) {
         const result = await mergeCompletedMilestone(basePath, mid);

package/dist/resources/extensions/gsd/parsers-legacy.js

@@ -165,16 +165,26 @@ function _parsePlanImpl(content) {
     const demo = extractBoldField(body, 'Demo') || '';
     const mhSection = extractSection(body, 'Must-Haves');
     const mustHaves = mhSection ? parseBullets(mhSection) : [];
+    // Parse tasks from ## Tasks section first, then scan the full body for any
+    // task checkboxes that were missed. Multi-task plans can interleave T01 detail
+    // headings (## Steps, ## Must-Haves) before T02's checkbox, which causes
+    // extractSection("Tasks") to stop at the first ## heading and miss T02+ (#3105).
     const tasksSection = extractSection(body, 'Tasks');
     const tasks = [];
-    if (tasksSection) {
-        const taskLines = tasksSection.split('\n');
+    // Parse task entries from a set of lines, appending to `tasks`.
+    const parseTaskLines = (lines, knownIds) => {
         let currentTask = null;
-        for (const line of taskLines) {
+        for (const line of lines) {
             const cbMatch = line.match(/^-\s+\[([ xX])\]\s+\*\*([\w.]+):\s+(.+?)\*\*\s*(.*)/);
             // Heading-style: ### T01 -- Title, ### T01: Title, ### T01 — Title
             const hdMatch = !cbMatch ? line.match(/^#{2,4}\s+([\w.]+)\s*(?:--|—|:)\s*(.+)/) : null;
             if (cbMatch || hdMatch) {
+                const taskId = cbMatch ? cbMatch[2] : hdMatch[1];
+                // Skip tasks already found in the Tasks section
+                if (knownIds.has(taskId)) {
+                    currentTask = null;
+                    continue;
+                }
                 if (currentTask)
                     tasks.push(currentTask);
                 if (cbMatch) {
@@ -229,7 +239,14 @@ function _parsePlanImpl(content) {
         }
         if (currentTask)
             tasks.push(currentTask);
+    };
+    if (tasksSection) {
+        parseTaskLines(tasksSection.split('\n'), new Set());
     }
+    // Second pass: scan the full body for task checkboxes outside ## Tasks.
+    // This handles interleaved plans where T02+ appear after T01's detail headings.
+    const foundIds = new Set(tasks.map(t => t.id));
+    parseTaskLines(body.split('\n'), foundIds);
     const filesSection = extractSection(body, 'Files Likely Touched');
     const filesLikelyTouched = filesSection ? parseBullets(filesSection) : [];
     const result = { id, title, goal, demo, mustHaves, tasks, filesLikelyTouched };

package/dist/resources/extensions/gsd/paths.js

@@ -289,11 +289,51 @@ export function gsdRoot(basePath) {
     gsdRootCache.set(basePath, result);
     return result;
 }
+/**
+ * Detect if a path is inside a .gsd/worktrees/<name>/ structure.
+ *
+ * GSD auto-worktrees live at <project>/.gsd/worktrees/<milestoneId>/.
+ * When gsdRoot() is called with such a path, we must NOT walk up to the
+ * project root's .gsd — each worktree manages its own .gsd state (#2594).
+ *
+ * Matches both forward-slash and platform-native separators to handle
+ * Windows paths (path.sep = '\\') and normalized Unix paths.
+ */
+function isInsideGsdWorktree(p) {
+    // Match /.gsd/worktrees/<name> where <name> is the final segment or
+    // followed by a separator. The <name> segment must be non-empty.
+    const sepFwd = "/";
+    const sepNative = "\\";
+    const markers = [
+        `${sepFwd}.gsd${sepFwd}worktrees${sepFwd}`,
+        `${sepNative}.gsd${sepNative}worktrees${sepNative}`,
+    ];
+    for (const marker of markers) {
+        const idx = p.indexOf(marker);
+        if (idx === -1)
+            continue;
+        // Verify there's a non-empty worktree name after the marker
+        const afterMarker = p.slice(idx + marker.length);
+        // The name is everything up to the next separator (or end of string)
+        const nameEnd = afterMarker.search(/[/\\]/);
+        const name = nameEnd === -1 ? afterMarker : afterMarker.slice(0, nameEnd);
+        if (name.length > 0)
+            return true;
+    }
+    return false;
+}
 function probeGsdRoot(rawBasePath) {
     // 1. Fast path — check the input path directly
     const local = join(rawBasePath, ".gsd");
     if (existsSync(local))
         return local;
+    // 1b. Worktree guard (#2594) — if basePath is inside a .gsd/worktrees/<name>/
+    //     structure, return the worktree-local .gsd path immediately. Without this,
+    //     the git-root probe (step 2) or walk-up (step 3) escapes to the project
+    //     root's .gsd, causing ensurePreconditions() and deriveState() to read/write
+    //     state in the wrong location.
+    if (isInsideGsdWorktree(rawBasePath))
+        return local;
     // Resolve symlinks so path comparisons work correctly across platforms
     // (e.g. macOS /var → /private/var). Use rawBasePath as fallback if not resolvable.
     let basePath;
@@ -303,6 +343,9 @@ function probeGsdRoot(rawBasePath) {
     catch {
         basePath = rawBasePath;
     }
+    // Also check the resolved path for the worktree pattern (macOS /tmp → /private/tmp)
+    if (basePath !== rawBasePath && isInsideGsdWorktree(basePath))
+        return local;
     // 2. Git root anchor — used as both probe target and walk-up boundary
     //    Only walk if we're inside a git project — prevents escaping into
     //    unrelated filesystem territory when running outside any repo.

package/dist/resources/extensions/gsd/preferences-models.js

@@ -53,6 +53,7 @@ export function resolveModelWithFallbacksForUnit(unitType) {
             break;
         case "complete-slice":
         case "complete-milestone":
+        case "worktree-merge":
         case "run-uat":
             phaseConfig = m.completion;
             break;
@@ -110,6 +111,18 @@ export function getNextFallbackModel(currentModelId, modelConfig) {
         return modelsToTry[0];
     }
 }
+/**
+ * Detect whether an error message indicates a transient network error
+ * (worth retrying the same model) vs a permanent provider error
+ * (auth failure, quota exceeded, etc. -- should fall back immediately).
+ */
+export function isTransientNetworkError(errorMsg) {
+    if (!errorMsg)
+        return false;
+    const hasNetworkSignal = /network|ECONNRESET|ETIMEDOUT|ECONNREFUSED|socket hang up|fetch failed|connection.*reset|dns/i.test(errorMsg);
+    const hasPermanentSignal = /auth|unauthorized|forbidden|invalid.*key|quota|billing/i.test(errorMsg);
+    return hasNetworkSignal && !hasPermanentSignal;
+}
 /**
  * Validate a model ID string.
  * Returns true if the ID looks like a valid model identifier.
@@ -273,7 +286,7 @@ export function resolveContextSelection() {
     return profile === "budget" ? "smart" : "full";
 }
 /**
- * Resolve the search provider preference from PREFERENCES.md.
+ * Resolve the search provider preference from preferences.md.
  * Returns undefined if not configured (caller falls back to existing behavior).
  */
 export function resolveSearchProviderFromPreferences() {

package/dist/resources/extensions/gsd/preferences-types.js

@@ -76,6 +76,7 @@ export const KNOWN_PREFERENCE_KEYS = new Set([
 export const KNOWN_UNIT_TYPES = [
     "research-milestone", "plan-milestone", "research-slice", "plan-slice",
     "execute-task", "reactive-execute", "gate-evaluate", "complete-slice", "replan-slice", "reassess-roadmap",
-    "run-uat", "complete-milestone",
+    "run-uat", "complete-milestone", "validate-milestone", "rewrite-docs",
+    "discuss-milestone", "discuss-slice", "worktree-merge",
 ];
 export const SKILL_ACTIONS = new Set(["use", "prefer", "avoid"]);

package/dist/resources/extensions/gsd/preferences.js

@@ -24,27 +24,27 @@ export { validatePreferences } from "./preferences-validation.js";
 // ─── Re-exports: skills ─────────────────────────────────────────────────────
 export { resolveAllSkillReferences, resolveSkillDiscoveryMode, resolveSkillStalenessDays, } from "./preferences-skills.js";
 // ─── Re-exports: models ─────────────────────────────────────────────────────
-export { resolveModelForUnit, resolveModelWithFallbacksForUnit, getNextFallbackModel, validateModelId, updatePreferencesModels, resolveDynamicRoutingConfig, resolveAutoSupervisorConfig, resolveProfileDefaults, resolveEffectiveProfile, resolveInlineLevel, resolveContextSelection, resolveSearchProviderFromPreferences, } from "./preferences-models.js";
+export { resolveModelForUnit, resolveModelWithFallbacksForUnit, getNextFallbackModel, isTransientNetworkError, validateModelId, updatePreferencesModels, resolveDynamicRoutingConfig, resolveAutoSupervisorConfig, resolveProfileDefaults, resolveEffectiveProfile, resolveInlineLevel, resolveContextSelection, resolveSearchProviderFromPreferences, } from "./preferences-models.js";
 // ─── Path Constants & Getters ───────────────────────────────────────────────
 function gsdHome() {
     return process.env.GSD_HOME || join(homedir(), ".gsd");
 }
 function globalPreferencesPath() {
-    return join(gsdHome(), "PREFERENCES.md");
+    return join(gsdHome(), "preferences.md");
 }
 function legacyGlobalPreferencesPath() {
     return join(homedir(), ".pi", "agent", "gsd-preferences.md");
 }
 function projectPreferencesPath() {
-    return join(gsdRoot(process.cwd()), "PREFERENCES.md");
+    return join(gsdRoot(process.cwd()), "preferences.md");
 }
-// Legacy: older versions used lowercase preferences.md.
-// Check lowercase as a fallback so those files aren't silently ignored.
-function globalPreferencesPathLegacy() {
-    return join(gsdHome(), "preferences.md");
+// Bootstrap in gitignore.ts historically created PREFERENCES.md (uppercase) by mistake.
+// Check uppercase as a fallback so those files aren't silently ignored.
+function globalPreferencesPathUppercase() {
+    return join(gsdHome(), "PREFERENCES.md");
 }
-function projectPreferencesPathLegacy() {
-    return join(gsdRoot(process.cwd()), "preferences.md");
+function projectPreferencesPathUppercase() {
+    return join(gsdRoot(process.cwd()), "PREFERENCES.md");
 }
 export function getGlobalGSDPreferencesPath() {
     return globalPreferencesPath();
@@ -58,12 +58,12 @@ export function getProjectGSDPreferencesPath() {
 // ─── Loading ────────────────────────────────────────────────────────────────
 export function loadGlobalGSDPreferences() {
     return loadPreferencesFile(globalPreferencesPath(), "global")
-        ?? loadPreferencesFile(globalPreferencesPathLegacy(), "global")
+        ?? loadPreferencesFile(globalPreferencesPathUppercase(), "global")
         ?? loadPreferencesFile(legacyGlobalPreferencesPath(), "global");
 }
 export function loadProjectGSDPreferences() {
     return loadPreferencesFile(projectPreferencesPath(), "project")
-        ?? loadPreferencesFile(projectPreferencesPathLegacy(), "project");
+        ?? loadPreferencesFile(projectPreferencesPathUppercase(), "project");
 }
 export function loadEffectiveGSDPreferences() {
     const globalPreferences = loadGlobalGSDPreferences();
@@ -149,7 +149,7 @@ export function parsePreferencesMarkdown(content) {
     }
     if (!_warnedUnrecognizedFormat) {
         _warnedUnrecognizedFormat = true;
-        console.warn("[parsePreferencesMarkdown] PREFERENCES.md exists but uses an unrecognized format — skipping.");
+        console.warn("[parsePreferencesMarkdown] preferences.md exists but uses an unrecognized format — skipping.");
     }
     return null;
 }
@@ -289,9 +289,6 @@ function mergePreferences(base, override) {
         service_tier: override.service_tier ?? base.service_tier,
         forensics_dedup: override.forensics_dedup ?? base.forensics_dedup,
         show_token_cost: override.show_token_cost ?? base.show_token_cost,
-        experimental: (base.experimental || override.experimental)
-            ? { ...(base.experimental ?? {}), ...(override.experimental ?? {}) }
-            : undefined,
     };
 }
 function mergeStringLists(base, override) {
@@ -412,7 +409,7 @@ export function resolvePreDispatchHooks() {
  * Resolve the effective git isolation mode from preferences.
  * Returns "none" (default), "worktree", or "branch".
  *
- * Default is "none" so GSD works out of the box without PREFERENCES.md.
+ * Default is "none" so GSD works out of the box without preferences.md.
  * Worktree isolation requires explicit opt-in because it depends on git
  * branch infrastructure that must be set up before use.
  */

package/dist/resources/extensions/gsd/prompt-loader.js

@@ -124,7 +124,10 @@ export function loadPrompt(name, vars = {}) {
         }
     }
     for (const [key, value] of Object.entries(effectiveVars)) {
-        content = content.replaceAll(`{{${key}}}`, value);
+        // Use split/join instead of replaceAll to avoid JavaScript's special
+        // replacement patterns ($', $`, $&) being interpreted in the value.
+        // See: https://github.com/gsd-build/gsd-2/issues/2968
+        content = content.split(`{{${key}}}`).join(value);
     }
     return content.trim();
 }

package/dist/resources/extensions/gsd/prompts/complete-milestone.md

@@ -56,7 +56,7 @@ Then:
    - `followUps` (string) — Follow-up items for future milestones
    - `deviations` (string) — Deviations from the original plan
 10. For each requirement whose status changed in step 8, call `gsd_requirement_update` with the requirement ID and updated `status` and `validation` fields — the tool regenerates `.gsd/REQUIREMENTS.md` automatically.
-11. Update `.gsd/PROJECT.md` to reflect milestone completion and current project state.
+11. Update `.gsd/PROJECT.md`: use the `write` tool with `path: ".gsd/PROJECT.md"` and `content` containing the full updated document reflecting milestone completion and current project state. Do NOT use the `edit` tool for this — PROJECT.md is a full-document refresh.
 12. Review all slice summaries for cross-cutting lessons, patterns, or gotchas that emerged during this milestone. Append any non-obvious, reusable insights to `.gsd/KNOWLEDGE.md`.
 13. Do not commit manually — the system auto-commits your changes after this unit completes.
 - Say: "Milestone {{milestoneId}} complete."

package/dist/resources/extensions/gsd/prompts/complete-slice.md

@@ -29,9 +29,11 @@ Then:
 8. Write `{{sliceUatPath}}` — a concrete UAT script with real test cases derived from the slice plan and task summaries. Include preconditions, numbered steps with expected outcomes, and edge cases. This must NOT be a placeholder or generic template — tailor every test case to what this slice actually built.
 9. Review task summaries for `key_decisions`. Append any significant decisions to `.gsd/DECISIONS.md` if missing.
 10. Review task summaries for patterns, gotchas, or non-obvious lessons learned. If any would save future agents from repeating investigation or hitting the same issues, append them to `.gsd/KNOWLEDGE.md`. Only add entries that are genuinely useful — don't pad with obvious observations.
-11. Call `gsd_complete_slice` with milestone_id, slice_id, the slice summary, and the UAT result. Do NOT manually mark the roadmap checkbox — the tool writes to the DB and renders the ROADMAP.md projection automatically.
+11. Call `gsd_complete_slice` with milestoneId, sliceId, the slice summary, and the UAT result. Do NOT manually mark the roadmap checkbox — the tool writes to the DB and renders the ROADMAP.md projection automatically.
 12. Do not run git commands — the system commits your changes and handles any merge after this unit succeeds.
-13. Update `.gsd/PROJECT.md` if it exists — refresh current state if needed.
+13. Update `.gsd/PROJECT.md` if it exists — refresh current state if needed: use the `write` tool with `path: ".gsd/PROJECT.md"` and `content` containing the full updated document reflecting current project state. Do NOT use the `edit` tool for this — PROJECT.md is a full-document refresh.
+
+**Autonomous execution:** Do not call `ask_user_questions` or `secure_env_collect`. You are running in auto-mode — there is no human available to answer questions. Make reasonable assumptions and document them in the slice summary. If a decision genuinely requires human input, note it in the summary and proceed with the best available option.
 
 **You MUST call `gsd_complete_slice` with the slice summary and UAT content before finishing. The tool persists to both DB and disk and renders `{{sliceSummaryPath}}` and `{{sliceUatPath}}` automatically.**
 

package/dist/resources/extensions/gsd/prompts/discuss-headless.md

@@ -38,7 +38,7 @@ Do a mandatory investigation pass before making any decisions. This is not optio
 3. **Web search** — `search-the-web` if the domain is unfamiliar, if you need current best practices, or if the spec references external services/APIs you need facts about. Use `fetch_page` for full content when snippets aren't enough.
 
 **Web search budget:** Budget carefully across investigation + focused research:
-- Prefer `resolve_library` / `get_library_docs` over `web_search` for library documentation.
+- Prefer `resolve_library` / `get_library_docs` over `search-the-web` for library documentation.
 - Prefer `search_and_read` for one-shot topic research.
 - Target 2-3 web searches in this investigation pass. Save remaining budget for focused research.
 - Do NOT repeat the same or similar queries.

package/dist/resources/extensions/gsd/prompts/discuss.md

@@ -37,7 +37,7 @@ Before asking your first question, do a mandatory investigation pass. This is no
 3. **Web search** — `search-the-web` if the domain is unfamiliar, if you need current best practices, or if the user referenced external services/APIs you need facts about. Use `fetch_page` for full content when snippets aren't enough.
 
 **Web search budget:** You have a limited number of web searches per turn (typically 3-5). The discuss phase spans many turns (investigation, question rounds, focused research, requirements), so budget carefully:
-- Prefer `resolve_library` / `get_library_docs` over `web_search` for library documentation — they don't consume the web search budget.
+- Prefer `resolve_library` / `get_library_docs` over `search-the-web` for library documentation — they don't consume the web search budget.
 - Prefer `search_and_read` for one-shot topic research — it combines search + page fetch in a single call.
 - Target 2-3 web searches in the investigation pass. Save remaining budget for the focused research pass before roadmap creation.
 - Do NOT repeat the same or similar queries. If a search didn't find what you need, rephrase once or move on.

package/dist/resources/extensions/gsd/prompts/execute-task.md

@@ -68,11 +68,13 @@ Then:
 17. If you discover a non-obvious rule, recurring gotcha, or useful pattern during execution, append it to `.gsd/KNOWLEDGE.md`. Only add entries that would save future agents from repeating your investigation. Don't add obvious things.
 18. Read the template at `~/.gsd/agent/extensions/gsd/templates/task-summary.md`
 19. Write `{{taskSummaryPath}}`
-20. Call `gsd_complete_task` with milestone_id, slice_id, task_id, and a summary of what was accomplished. This is your final required step — do NOT manually edit PLAN.md checkboxes. The tool marks the task complete, updates the DB, and renders PLAN.md automatically.
+20. Call `gsd_complete_task` with milestoneId, sliceId, taskId, and a summary of what was accomplished. This is your final required step — do NOT manually edit PLAN.md checkboxes. The tool marks the task complete, updates the DB, and renders PLAN.md automatically.
 21. Do not run git commands — the system reads your task summary after completion and creates a meaningful commit from it (type inferred from title, message from your one-liner, key files from frontmatter). Write a clear, specific one-liner in the summary — it becomes the commit message.
 
 All work stays in your working directory: `{{workingDirectory}}`.
 
+**Autonomous execution:** Do not call `ask_user_questions` or `secure_env_collect`. You are running in auto-mode — there is no human available to answer questions. Make reasonable assumptions and document them in the task summary. If a decision genuinely requires human input, note it in the summary and proceed with the best available option.
+
 **You MUST call `gsd_complete_task` AND write `{{taskSummaryPath}}` before finishing.**
 
 When done, say: "Task {{taskId}} complete."

package/dist/resources/extensions/gsd/prompts/forensics.md

@@ -102,6 +102,8 @@ A stale lock (PID is dead) means the previous auto-mode session crashed mid-unit
 
 A unit dispatched more than once (`type/id` appears multiple times) indicates a stuck loop — the unit completed but artifact verification failed.
 
+{{dedupSection}}
+
 ## Investigation Protocol
 
 1. **Start with the pre-parsed forensic report** above. The anomaly section contains automated findings — treat these as leads, not conclusions.
@@ -133,8 +135,6 @@ Explain your findings:
 - **Code snippet** — the problematic code and what it should do instead
 - **Recovery** — what the user can do right now to get unstuck
 
-{{dedupSection}}
-
 Then **offer GitHub issue creation**: "Would you like me to create a GitHub issue for this on gsd-build/gsd-2?"
 
 **CRITICAL: The `github_issues` tool ONLY targets the current user's repository — it has no `repo` parameter. You MUST use `gh issue create --repo gsd-build/gsd-2` via the `bash` tool to file on the correct repo. Do NOT use the `github_issues` tool for this.**

package/dist/resources/extensions/gsd/prompts/guided-discuss-milestone.md

@@ -13,7 +13,7 @@ Discuss milestone {{milestoneId}} ("{{milestoneTitle}}"). Identify gray areas, a
 Do a lightweight targeted investigation so your questions are grounded in reality:
 - Scout the codebase (`rg`, `find`, or `scout`) to understand what already exists that this milestone touches or builds on
 - Check the roadmap context above (if present) to understand what surrounds this milestone
-- Use `resolve_library` / `get_library_docs` for unfamiliar libraries — prefer this over `web_search` for library documentation
+- Use `resolve_library` / `get_library_docs` for unfamiliar libraries — prefer this over `search-the-web` for library documentation
 - Identify the 3–5 biggest behavioural and architectural unknowns: things where the user's answer will materially change what gets built
 
 **Web search budget:** You have a limited number of web searches per turn (typically 3-5). Prefer `resolve_library` / `get_library_docs` for library documentation and `search_and_read` for one-shot topic research — they are more budget-efficient. Target 2-3 web searches in the investigation pass. Distribute remaining searches across subsequent question rounds rather than clustering them.

package/dist/resources/extensions/gsd/prompts/guided-discuss-slice.md

@@ -13,7 +13,7 @@ Your goal is **not** to center the discussion on tech stack trivia, naming conve
 Do a lightweight targeted investigation so your questions are grounded in reality:
 - Scout the codebase (`rg`, `find`, or `scout` for broad unfamiliar areas) to understand what already exists that this slice touches or builds on
 - Check the roadmap context above to understand what surrounds this slice — what comes before, what depends on it
-- Use `resolve_library` / `get_library_docs` for unfamiliar libraries — prefer this over `web_search` for library documentation
+- Use `resolve_library` / `get_library_docs` for unfamiliar libraries — prefer this over `search-the-web` for library documentation
 - Identify the 3–5 biggest behavioural unknowns: things where the user's answer will materially change what gets built
 
 **Web search budget:** You have a limited number of web searches per turn (typically 3-5). Prefer `resolve_library` / `get_library_docs` for library documentation and `search_and_read` for one-shot topic research — they are more budget-efficient. Target 2-3 web searches in the investigation pass. Distribute remaining searches across subsequent question rounds rather than clustering them.

package/dist/resources/extensions/gsd/prompts/plan-slice.md

@@ -82,6 +82,8 @@ Then:
 
 The slice directory and tasks/ subdirectory already exist. Do NOT mkdir. All work stays in your working directory: `{{workingDirectory}}`.
 
+**Autonomous execution:** Do not call `ask_user_questions` or `secure_env_collect`. You are running in auto-mode — there is no human available to answer questions. Make reasonable assumptions and document them in the plan. If a decision genuinely requires human input, write a note in the relevant task's description and call `gsd_plan_slice` with what you have.
+
 **You MUST call `gsd_plan_slice` to persist the planning state before finishing.**
 
 When done, say: "Slice {{sliceId}} planned."

package/dist/resources/extensions/gsd/prompts/rethink.md

@@ -80,4 +80,4 @@ If a proposed order would violate constraints, explain the issue and suggest alt
 - Do NOT park completed milestones — it would corrupt dependency satisfaction
 - Park is preferred over discard when a milestone has any completed work
 - Always persist queue order changes to `.gsd/QUEUE-ORDER.json`
-- After changes, run `git add .gsd/ && git commit -m "docs(gsd): rethink milestone plan"` to persist (rethink runs interactively outside auto-mode, so no system auto-commit)
+- {{commitInstruction}}