⚡ Bolt: Optimize bulk worktree creation by hoisting loadConfig

.jules/bolt.md

@@ -4,3 +4,6 @@
 ## 2026-05-16 - [Batch FeathersJS user fetches with $in operator to fix N+1 query]
 **Learning:** FeathersJS allows passing `$in` clauses through the query parameter (e.g. `user_id: { $in: ownerIds }`). When writing custom Feathers service logic, you can easily parse this array and pass it to Drizzle's `inArray()` to perform a batched query, instead of looping over `service.get(id)` causing N+1 database roundtrips.
 **Action:** When implementing or updating custom Feathers `find()` methods, extract and parse the `$in` parameters to support batched Drizzle `inArray()` lookups, and always replace `Promise.all(ids.map(id => service.get(id)))` with a single batched `find()` call.
+## 2025-02-12 - [Optimize bulk worktree creation by hoisting loadConfig]
+**Learning:** When performing bulk array mapping operations inside a class method (e.g. `create` taking `data[]`), repeatedly calling internal instance methods that perform synchronous disk I/O or configuration parsing (like `loadConfig()`) can easily trigger performance degradation via N simultaneous file reads.
+**Action:** When mapping over an array to process records and an underlying shared configuration or static data is needed, evaluate whether it can be loaded once *before* the mapping operation and explicitly passed into the individual worker method to prevent O(N) configuration parsing/disk I/O.

apps/agor-daemon/src/services/worktrees.ts

@@ -10,7 +10,13 @@ import { existsSync } from 'node:fs';
 import { mkdir } from 'node:fs/promises';
 import { homedir } from 'node:os';
 import { dirname, join } from 'node:path';
-import { ENVIRONMENT, isWorktreeRbacEnabled, loadConfig, PAGINATION } from '@agor/core/config';
+import {
+  type AgorConfig,
+  ENVIRONMENT,
+  isWorktreeRbacEnabled,
+  loadConfig,
+  PAGINATION,
+} from '@agor/core/config';
 import { type Database, WorktreeRepository, type WorktreeWithZoneAndSessions } from '@agor/core/db';
 import { renderWorktreeSnapshot } from '@agor/core/environment/render-snapshot';
 import type { Application } from '@agor/core/feathers';
@@ -226,8 +232,11 @@ export class WorktreesService extends DrizzleService<Worktree, Partial<Worktree>
    * so admins can set org-wide defaults in config.yaml. Explicit values on the
    * input always win; defaults fill in only when the caller omits the field.
    */
-  private async applyWorktreeCreateDefaults(data: Partial<Worktree>): Promise<Partial<Worktree>> {
-    const config = await loadConfig();
+  private async applyWorktreeCreateDefaults(
+    data: Partial<Worktree>,
+    preloadedConfig?: AgorConfig
+  ): Promise<Partial<Worktree>> {
+    const config = preloadedConfig ?? (await loadConfig());
     const defaults = config.worktrees;
     if (!defaults) return data;
 
@@ -252,8 +261,11 @@ export class WorktreesService extends DrizzleService<Worktree, Partial<Worktree>
     params?: WorktreeParams
   ): Promise<Worktree | Worktree[]> {
     if (Array.isArray(data)) {
+      // ⚡ Bolt Performance Optimization:
+      // Load config once before the loop to prevent N simultaneous disk I/O reads.
+      const config = await loadConfig();
       const withDefaults = await Promise.all(
-        data.map((item) => this.applyWorktreeCreateDefaults(item))
+        data.map((item) => this.applyWorktreeCreateDefaults(item, config))
       );
       return super.create(withDefaults, params) as Promise<Worktree[]>;
     }

plan.md

@@ -0,0 +1,22 @@
+1. **Identify the Bottleneck**:
+   - `loadConfig()` reads and parses the `~/.agor/config.yaml` file on every single call.
+   - In `apps/agor-daemon/src/services/worktrees.ts`, the `create` method processes bulk inputs using `Promise.all(data.map(item => this.applyWorktreeCreateDefaults(item)))`.
+   - Since `applyWorktreeCreateDefaults` internally calls `await loadConfig()`, this triggers N simultaneous disk I/O reads (N = number of items). This can cause major performance issues when processing large arrays.
+
+2. **Implement the Optimization**:
+   - Instead of calling `await loadConfig()` inside `applyWorktreeCreateDefaults` (which gets called in the map loop), we can load the config **once** outside the loop in the `create` method.
+   - Update `applyWorktreeCreateDefaults` to accept the loaded `config` as an optional parameter (or always pass it from `create`).
+   - If not passed, default to loading it once (to support other places that might use it).
+   - We will modify `apps/agor-daemon/src/services/worktrees.ts`.
+
+3. **Verify**:
+   - Run tests for `@agor/daemon`.
+   - Run linter/formatter.
+
+4. **Pre-commit**:
+   - Complete pre commit steps to ensure proper testing, verification, review, and reflection are done.
+
+5. **Create PR**:
+   - Branch: `perf/worktree-bulk-create`
+   - Title: `⚡ Bolt: Optimize bulk worktree creation by hoisting loadConfig`
+   - PR description explaining the N disk I/O reduction.