personal-ai 0.1.0

package/README.md

@@ -0,0 +1,227 @@
+# pai — Personal AI Identity Provider
+
+本地优先的 AI Agent 身份画像系统。一条命令扫描本机，编译你的 profile，部署到任何 Agent。
+
+```
+pai init → Scan → Profile → Deploy to Agents
+```
+
+**Profile ≠ Skill**: Profile 描述你是谁（身份、环境、偏好、项目），Skill 是 Agent 使用 pai 的说明书。
+
+## 快速开始
+
+```bash
+# 安装 (需要 Node.js >= 22)
+npm install -g personal-ai
+
+# 一条命令完成: 初始化 + 扫描本机 + 编译 profile (~12s, 零 LLM)
+pai init
+
+# 部署到 Cursor (Agent 自动认识你)
+pai distribute
+
+# 查看你的 profile
+pai profile
+```
+
+三步搞定，不需要 LLM，不需要 API Key。Agent 立刻认识你。
+
+### 从源码安装 (开发者)
+
+```bash
+git clone https://github.com/piai/personal-ai-skills-new.git
+cd personal-ai-skills-new
+pnpm install && pnpm build && npm link
+```
+
+## 核心概念
+
+### Profile (核心产物)
+
+`~/.pai/profile.md` — 从本机扫描数据直接编译，零 LLM 依赖。
+
+包含 7 个维度:
+- **Identity**: 用户名、Apple ID、Git 身份、语言、时区
+- **Environment & Tools**: 运行时、包管理器、Shell、IDE 扩展
+- **Work Style & Habits**: 命令习惯、编码规则、AI agent 配置
+- **Active Projects & Recent Focus**: 活跃仓库、近期 commit、工作目录
+- **Digital Footprint**: 书签、浏览域名、安装的应用
+- **Registry & Cloud Accounts**: npm/Docker/AWS/GCP/Vercel 等账户
+- **Context**: 日历、文件组织结构
+
+### 14 个数据收集器
+
+| # | Collector | 数据来源 |
+|---|---|---|
+| 1 | identity-profile | 用户名、Apple ID、Git 配置、语言、时区 |
+| 2 | calendar-context | Calendar.app 订阅 |
+| 3 | file-organization | Documents/Desktop/Downloads 结构 |
+| 4 | dev-environment | 运行时版本、包管理器、全局包 |
+| 5 | dev-preferences | Shell aliases、Git config、Cursor 扩展 |
+| 6 | shell-habits | 最常用命令 Top 30 |
+| 7 | coding-rules | CLAUDE.md、Claude commands、项目规则 |
+| 8 | active-projects | 活跃 Git 仓库、SSH hosts |
+| 9 | productivity-setup | 已装 App、Dock 应用、浏览器 |
+| 10 | browser-bookmarks | 书签文件夹结构 |
+| 11 | browser-domains | 近 30 天高频域名 Top 30 |
+| 12 | github-profile | GitHub 用户信息、仓库、stars |
+| 13 | recent-focus | 近期 commit 分析、工作目录、工具使用 |
+| 14 | social-profiles | npm/Docker/AWS/GCP/Vercel 账户 |
+
+### Skill (Agent 说明书)
+
+`pai distribute` 部署到 Agent 的文件包含两部分:
+1. 你的 Profile (谁、环境、偏好)
+2. Agent Skill 指令 (如何调用 `pai context/search/add`)
+
+## CLI 命令
+
+```bash
+# Profile (核心)
+pai init                           # 初始化 + 扫描 + 编译 profile (一条龙)
+pai init --skip-scan               # 仅初始化，跳过扫描 (CI/测试用)
+pai profile                        # 查看当前 profile
+pai profile --rebuild              # 重新扫描 + 编译
+pai profile --export               # 输出可复制粘贴的 profile
+pai profile --json                 # JSON 元数据
+pai distribute                     # 部署 profile + skill 到 Cursor
+
+# 数据管理 (进阶)
+pai reset [--force]                # 清空所有数据并重新初始化
+pai add <text>                     # 手动添加文本到 raw/local/
+pai add --url <url>                # 抓取 URL 到 raw/web/
+pai add <file>                     # 添加文件内容
+pai auth google                    # Google 授权 (gmail/calendar 首次或重授权)
+pai import --source mac            # 手动触发 Mac 扫描 (写入 raw)
+pai import --source gmail [--days N] [--query "..."]   # Gmail 导入 (未授权时自动弹窗)
+pai import --source calendar [--days N]                # 日历导入
+pai import --source <src> --path   # 批量导入其他数据源 (需 --path)
+
+# 深度知识 (可选，需要 LLM)
+pai distill [--dry-run] [--file]   # 蒸馏 raw → vault (需 OPENAI_API_KEY)
+pai generate [--profile <name>]    # LLM 生成 SKILL.md
+pai index                          # 更新 QMD 索引
+
+# 搜索 & 检索
+pai ask <question>                 # 智能问答，端到端答案 (agent 首选，需 LLM)
+pai ask <question> --json          # JSON: answer, sources, steps
+pai search <query>                 # 混合搜索 vault
+pai search <query> --fast          # 快速关键词搜索
+pai search <query> --json          # JSON 格式
+pai context --task <desc>          # 身份 + 任务相关记忆 (agent 用)
+
+# 状态
+pai status [--json]                # 数据状态概览
+```
+
+## 技术栈
+
+- **语言**: TypeScript (ESM, strict mode)
+- **运行时**: Node.js ≥ 22
+- **CLI 框架**: Commander.js
+- **配置验证**: Zod + JSON5
+- **LLM 调用**: openai SDK (可选，深度知识功能需要)
+- **本地搜索**: QMD (可选，搜索功能需要)
+- **终端输出**: chalk + ora
+- **构建**: tsdown
+- **测试**: Vitest
+- **Lint**: Oxlint
+- **包管理**: pnpm
+
+## 前置依赖
+
+```bash
+# QMD (本地搜索引擎，搜索功能需要，profile 不需要)
+npm install -g https://github.com/tobi/qmd
+
+# GitHub CLI (可选，GitHub profile 收集)
+brew install gh && gh auth login
+```
+
+## 项目结构
+
+```
+src/
+├── cli/           # CLI 命令注册 (Commander.js)
+│   ├── build-program.ts
+│   ├── command-registry.ts
+│   └── register.*.ts
+├── profile/       # Profile 编译器 (scan → profile.md, 零 LLM)
+│   ├── compile.ts
+│   └── index.ts
+├── config/        # 配置管理 (Zod + JSON5)
+├── auth/          # Google OAuth (encryption + google-oauth)
+├── connectors/    # 数据收集器
+│   ├── mac/       # 14 个 Mac collectors
+│   ├── google/    # gmail.ts, calendar.ts
+│   └── sanitize.ts
+├── raw/           # Raw 层
+├── scraper/       # 网页抓取
+├── distill/       # 蒸馏 Pipeline (可选)
+├── generate/      # SKILL.md 生成 (可选)
+├── search/        # QMD 搜索封装
+├── llm/           # OpenAI client
+├── prompts/       # Prompt 模板
+├── utils/         # 工具函数
+├── types.ts       # 全局类型
+├── index.ts       # Public API
+└── entry.ts       # CLI 入口
+```
+
+## 数据目录
+
+```
+~/.pai/
+├── profile.md            # 核心产物 — 编译后的用户画像
+├── raw/                  # 原始数据 (扫描/添加)
+│   ├── local/
+│   ├── web/
+│   └── connector/       # mac/, gmail/, calendar/
+├── credentials/          # Google OAuth: client_secret.json, google-oauth.json.enc
+├── vault/                # 蒸馏后知识 (可选进阶)
+├── skills/profiles/      # LLM 生成的 SKILL.md (可选)
+└── config/
+    ├── pai.json5
+    ├── profiles.json5
+    └── preferences.md
+```
+
+## Agent 集成
+
+### 两层记忆架构
+
+```
+Layer 1 (常驻): pai distribute → ~/.cursor/rules/pai-context.mdc
+  → Profile (谁) + Skill (怎么用 pai), 每次对话自动注入
+
+Layer 2 (情境): pai ask "问题" → 直接答案 (agentic，需 LLM)
+            或 pai context --task "当前任务" → 身份 + 相关记忆
+```
+
+### 快速部署
+
+```bash
+pai init && pai distribute    # 两步搞定
+
+# Agent 在工作时调用
+pai ask "What's the user's deployment preference?"   # 推荐：直接拿答案
+pai context --task "configure PostgreSQL connection pooling"
+pai search "database performance" --json
+pai add "learned: always set pool_size=20 for production"
+```
+
+## 开发
+
+```bash
+pnpm install          # 安装依赖
+pnpm dev              # 开发模式 (tsx)
+pnpm build            # 构建 (tsdown)
+pnpm lint             # Lint (oxlint)
+pnpm typecheck        # 类型检查
+pnpm test             # 测试 (vitest)
+pnpm check            # 完整门禁
+```
+
+## License
+
+MIT

package/SKILL.md

@@ -0,0 +1,310 @@
+# pai — Personal AI Identity Provider
+
+> One command to scan your machine, compile your profile, and deploy to any AI agent.
+
+<!--
+metadata:
+  {
+    "name": "pai",
+    "version": "0.1.0",
+    "description": "Local-first AI agent identity system. Scan, compile profile, deploy to agents.",
+    "requires": {
+      "bins": ["pai"],
+      "optional_bins": ["qmd"],
+      "env": [],
+      "optional_env": ["OPENAI_API_KEY"]
+    },
+    "permissions": ["read", "write", "search"],
+    "homepage": "https://github.com/piai/personal-ai-skills-new"
+  }
+-->
+
+## What This Skill Does
+
+pai is a local-first AI agent identity provider. It scans your machine, compiles a personal profile, and deploys it to AI agents. When you have access to this skill, you can:
+
+- **Know** who the user is — identity, environment, tools, projects, preferences
+- **Search** the user's personal knowledge base (vault + raw)
+- **Ask** questions about the user and get direct answers (agentic)
+- **Remember** new lessons, preferences, and discoveries for the user
+
+## Quick Setup (1 minute)
+
+```bash
+npm install -g personal-ai   # Install pai CLI
+pai init                      # Scan machine + compile profile (~12s, no LLM needed)
+pai distribute                # Deploy to Cursor / Claude Code / agents
+```
+
+That's it. Your agent now knows who you are.
+
+## Authorization
+
+### Required Environment
+
+| Requirement | How to Check | How to Get |
+|---|---|---|
+| `pai` CLI | `pai --version` | `npm install -g personal-ai` |
+| Node.js ≥ 22 | `node -v` | `fnm install 22` or [nodejs.org](https://nodejs.org) |
+
+### Optional (for advanced features)
+
+| Requirement | Feature | How to Get |
+|---|---|---|
+| `qmd` CLI | Search (`pai search`) | `npm install -g https://github.com/tobi/qmd` |
+| `OPENAI_API_KEY` | AI features (`pai ask`, `pai distill`) | Set in shell profile |
+| Google OAuth | Gmail/Calendar import | `pai auth google` (built-in flow) |
+
+### Data Location
+
+All data is stored locally at `~/.pai/` (override with `PAI_HOME` env var):
+
+```
+~/.pai/
+├── profile.md    # Core output — compiled user profile (no LLM needed)
+├── raw/          # Original input (immutable after write)
+│   ├── local/    # Text & file input
+│   ├── web/      # Scraped URLs
+│   └── connector/# Imported data (mac scan, gmail, calendar)
+├── vault/        # Distilled knowledge (living documents, optional)
+├── credentials/  # Google OAuth tokens (encrypted)
+├── skills/profiles/ # LLM-generated SKILL.md files (optional)
+└── config/       # pai.json5 + profiles.json5 + preferences.md
+```
+
+### Verifying Access
+
+```bash
+pai --version      # Check pai is installed
+pai status         # Check data directory and counts
+pai profile        # View your profile
+```
+
+If `pai` is not found:
+```bash
+npm install -g personal-ai
+pai init
+```
+
+## Available Commands
+
+### Adding Knowledge
+
+```bash
+# Add a text experience/lesson
+pai add "Always use connection pooling with PostgreSQL in production"
+
+# Add content from a URL (scrapes and saves)
+pai add --url "https://docs.example.com/best-practices"
+
+# Add a local file
+pai add ./meeting-notes.txt
+
+# Import from Mac system scan (no --path; auto-scans 11 dimensions)
+pai import --source mac
+pai import --source mac --dry-run
+
+# Import from a data connector directory
+pai import --source gmail --path ~/data/gmail-export/
+```
+
+### Ask (RECOMMENDED for Agents — get direct answers)
+
+Ask a question; an agentic secretary uses tools to find the answer and returns a direct reply. No need to interpret raw search results.
+
+**Requires:** `OPENAI_API_KEY` (agentic loop uses LLM + tools).
+
+**Tools available to the agent:** search vault/raw, read profile, read file, grep (ripgrep), glob, ls, bash. The agent decides which to call and when to stop.
+
+```bash
+# Ask anything about the user
+pai ask "What deployment method does this user prefer?"
+pai ask "Does the user have Kubernetes experience?" --json
+
+# Options
+pai ask "用户的编码规范是什么？" --steps 15   # max tool-call steps (default: 10)
+pai ask "..." --model gpt-4o                   # override model
+pai ask "..." --verbose                        # show each tool step
+```
+
+**Output:** plain answer; or with `--json`: `{ "answer", "sources", "steps" }`.
+
+### Retrieving Context (fast, no LLM)
+
+```bash
+# Get identity + task-relevant vault memories (hybrid search)
+pai context --task "current task description"
+
+# Get identity only (no search, instant)
+pai context
+
+# Machine-readable output
+pai context --task "deploy React app" --json
+
+# Use a specific profile
+pai context --profile coding-assistant --task "..."
+```
+
+### Searching Knowledge
+
+```bash
+# Search distilled knowledge (vault)
+pai search "PostgreSQL performance"
+
+# Search raw original data (for tracing back)
+pai search "PostgreSQL" --raw
+
+# Search everything
+pai search "PostgreSQL" --all
+
+# Control result count
+pai search "React hooks" -n 10
+
+# Machine-readable output (for agents)
+pai search "PostgreSQL" --json
+```
+
+### Processing & Distilling
+
+```bash
+# Preview what would be distilled (safe, no writes)
+pai distill --dry-run
+
+# Process all pending raw files → extract to vault
+pai distill
+
+# Process a specific file
+pai distill --file ~/.pai/raw/local/2026-02-06T15-17-xxx.md
+
+# Update QMD search index after changes
+pai index
+```
+
+### Generating SKILL.md
+
+```bash
+# Generate all configured profiles
+pai generate
+
+# Generate a specific profile
+pai generate --profile coding-assistant
+
+# View generated profile
+cat ~/.pai/skills/profiles/coding-assistant.md
+```
+
+### Agent Deployment
+
+```bash
+# Deploy identity + agent instructions to Cursor rules and other agent configs
+pai distribute                             # Default: all targets
+pai distribute --target cursor             # Cursor only
+pai distribute --target claude             # Claude Code only
+pai distribute --profile coding-assistant  # Use a specific generated profile
+```
+
+### Status & Reset
+
+```bash
+# Overview of all data
+pai status
+pai status --json         # Machine-readable
+
+# Remove all data and re-initialize (clean slate for testing)
+pai reset           # Prompts for confirmation
+pai reset --force   # No prompt, immediate wipe + init
+```
+
+## How to Use This Skill as an Agent
+
+### 1. Ask questions (recommended)
+
+For any question about the user, use `pai ask` to get a direct answer. The agent uses tools (vault search, profile, grep, bash) and returns a concise reply.
+
+```bash
+pai ask "What does this user prefer for deployment?"
+pai ask "What's the user's current project?" --json
+```
+
+### 2. Retrieve context before starting work
+
+For a quick identity + task-relevant snippets (no LLM, or light):
+
+```bash
+pai context --task "brief description of what you're about to do"
+```
+
+When you need raw search results (chunks) instead of a synthesized answer:
+
+```bash
+pai search "editor preferences"
+pai search "React deployment" --json -n 3
+```
+
+When you discover something the user should remember (a lesson, preference, or tip):
+
+```bash
+pai add "The user prefers dark mode and monospace fonts in all editors"
+pai add --url "https://the-useful-article.com"
+```
+
+### 5. Periodic maintenance
+
+After adding multiple items, process and regenerate:
+
+```bash
+pai distill     # Process pending raw files into vault
+pai generate    # Regenerate SKILL.md profiles
+pai distribute  # Update deployed agent configs
+```
+
+The generated profiles at `~/.pai/skills/profiles/` can be used by other agents.
+
+## Data Flow
+
+```
+User Input ──→ pai add ──→ raw/     (original, immutable)
+                              │
+                              ▼
+                         pai distill ──→ vault/  (structured knowledge, 1:N routing)
+                              │
+                              ▼
+                         pai generate ──→ skills/profiles/*.md
+                              │
+                              ▼
+                         pai distribute ──→ ~/.cursor/rules/ (auto-injected)
+                              │
+                              ▼
+Agent starts ──→ reads identity from rules (Layer 1: passive)
+Agent works  ──→ pai ask "question" (Layer 2: direct answer) or pai context --task "..." (quick retrieval)
+Agent learns ──→ pai add "new lesson" (Layer 3: write-back)
+```
+
+## Security Notes
+
+- All data stays on the user's local machine (`~/.pai/`)
+- No data is sent anywhere except to the configured LLM API for processing
+- `OPENAI_API_KEY` is read from environment, never stored in files
+- Raw files are immutable after creation (append-only log)
+- The user controls what goes in and what gets distilled
+
+## Profiles Configuration
+
+Profiles are defined in `~/.pai/config/profiles.json5`. Each profile specifies which vault directories to include and the max output size:
+
+```json5
+{
+  profiles: {
+    "coding-assistant": {
+      scope: ["vault/coding/**", "vault/preferences/coding-style.md"],
+      maxLines: 30,
+    },
+    "full-context": {
+      scope: ["vault/**"],
+      maxLines: 50,
+    },
+  },
+}
+```
+
+To add a new profile, edit `profiles.json5` and run `pai generate`.

package/dist/auth-Dtx8Wc3l.mjs

@@ -0,0 +1,2 @@
+#!/usr/bin/env node
+export {  };

package/dist/calendar-BHcM4wfQ.mjs

@@ -0,0 +1,91 @@
+#!/usr/bin/env node
+import { a as info, c as warn, r as googleOAuth } from "./entry.mjs";
+import "./auth-Dtx8Wc3l.mjs";
+import { google } from "googleapis";
+
+//#region src/connectors/google/calendar.ts
+/**
+* Google Calendar connector — fetches events and returns CollectorResult[] for raw layer.
+*
+* Default behavior: enumerate ALL user-visible calendars via calendarList.list(),
+* tag each event with the calendar display name so the ask agent can distinguish
+* "Work" meetings from "Holiday" markers.
+*/
+function formatEventTime(start) {
+	const dt = start.dateTime ?? start.date;
+	const tz = start.timeZone ?? "UTC";
+	if (!dt) return "";
+	return `${dt} (${tz})`;
+}
+/** Resolve calendar IDs + display names. If explicit list given, use as-is; otherwise enumerate all. */
+async function resolveCalendars(calendarApi, explicitIds) {
+	if (explicitIds) return explicitIds.map((id) => ({
+		id,
+		name: id
+	}));
+	return ((await calendarApi.calendarList.list({ showHidden: false })).data.items ?? []).filter((c) => c.id).map((c) => ({
+		id: c.id,
+		name: c.summary ?? c.id
+	}));
+}
+/**
+* Sync Google Calendar events and return one CollectorResult per event.
+* Each event includes a **Calendar** field with the display name of its source calendar.
+*/
+async function syncCalendar(opts = {}) {
+	await googleOAuth.ensureAuthenticated();
+	const calendarApi = google.calendar({
+		version: "v3",
+		auth: googleOAuth.getClient()
+	});
+	const calendars = await resolveCalendars(calendarApi, opts.calendars);
+	const lookbackDays = opts.lookbackDays ?? 30;
+	const lookforwardDays = opts.lookforwardDays ?? 90;
+	const now = /* @__PURE__ */ new Date();
+	const timeMin = (/* @__PURE__ */ new Date(now.getTime() - lookbackDays * 24 * 60 * 60 * 1e3)).toISOString();
+	const timeMax = new Date(now.getTime() + lookforwardDays * 24 * 60 * 60 * 1e3).toISOString();
+	info(`Fetching events from ${calendars.length} calendar(s): ${calendars.map((c) => c.name).join(", ")}`);
+	const results = [];
+	let failed = 0;
+	for (const cal of calendars) try {
+		const items = (await calendarApi.events.list({
+			calendarId: cal.id,
+			timeMin,
+			timeMax,
+			singleEvents: true,
+			orderBy: "startTime",
+			maxResults: 250
+		})).data.items ?? [];
+		for (const event of items) {
+			const id = event.id ?? "";
+			const summary = event.summary ?? "(no title)";
+			const startStr = event.start ? formatEventTime(event.start) : "";
+			const endStr = event.end ? formatEventTime(event.end) : "";
+			const location = event.location ?? "";
+			const attendees = (event.attendees ?? []).map((a) => a.email ?? a.displayName ?? "").filter(Boolean).join(", ");
+			const status = event.status ?? "confirmed";
+			const lines = [
+				`- **Calendar**: ${cal.name}`,
+				startStr ? `- **Time**: ${startStr}${endStr ? ` - ${endStr}` : ""}` : "",
+				location ? `- **Location**: ${location}` : "",
+				attendees ? `- **Attendees**: ${attendees}` : "",
+				`- **Status**: ${status}`
+			].filter(Boolean);
+			results.push({
+				id,
+				title: summary,
+				content: lines.join("\n") || "(no details)"
+			});
+		}
+	} catch (err) {
+		failed++;
+		const msgText = err instanceof Error ? err.message : String(err);
+		warn(`Calendar fetch failed for ${cal.name} (${cal.id}): ${msgText}`);
+	}
+	if (failed > 0 && calendars.length > failed) warn(`Calendar fetch failed for ${failed} of ${calendars.length} calendar(s).`);
+	return results;
+}
+
+//#endregion
+export { syncCalendar };
+//# sourceMappingURL=calendar-BHcM4wfQ.mjs.map

package/dist/calendar-BHcM4wfQ.mjs.map

@@ -0,0 +1 @@
+{"version":3,"file":"calendar-BHcM4wfQ.mjs","names":[],"sources":["../src/connectors/google/calendar.ts"],"sourcesContent":["/**\n * Google Calendar connector — fetches events and returns CollectorResult[] for raw layer.\n *\n * Default behavior: enumerate ALL user-visible calendars via calendarList.list(),\n * tag each event with the calendar display name so the ask agent can distinguish\n * \"Work\" meetings from \"Holiday\" markers.\n */\n\nimport { google } from \"googleapis\";\nimport type { calendar_v3 } from \"googleapis\";\nimport type { CollectorResult } from \"../../types.js\";\nimport { googleOAuth } from \"../../auth/index.js\";\nimport * as console from \"../../utils/console.js\";\n\nexport interface SyncCalendarOptions {\n  /** Calendar IDs to fetch. If omitted, fetches ALL user-visible calendars. */\n  calendars?: string[];\n  /** Days to look back (default 30). */\n  lookbackDays?: number;\n  /** Days to look forward (default 90). */\n  lookforwardDays?: number;\n}\n\nfunction formatEventTime(start: calendar_v3.Schema$EventDateTime): string {\n  const dt = start.dateTime ?? start.date;\n  const tz = start.timeZone ?? \"UTC\";\n  if (!dt) return \"\";\n  return `${dt} (${tz})`;\n}\n\n/** Resolve calendar IDs + display names. If explicit list given, use as-is; otherwise enumerate all. */\nasync function resolveCalendars(\n  calendarApi: calendar_v3.Calendar,\n  explicitIds?: string[],\n): Promise<{ id: string; name: string }[]> {\n  if (explicitIds) {\n    return explicitIds.map((id) => ({ id, name: id }));\n  }\n\n  // Enumerate all user-visible calendars\n  const res = await calendarApi.calendarList.list({ showHidden: false });\n  const items = res.data.items ?? [];\n  return items\n    .filter((c) => c.id)\n    .map((c) => ({\n      id: c.id!,\n      name: c.summary ?? c.id!,\n    }));\n}\n\n/**\n * Sync Google Calendar events and return one CollectorResult per event.\n * Each event includes a **Calendar** field with the display name of its source calendar.\n */\nexport async function syncCalendar(\n  opts: SyncCalendarOptions = {},\n): Promise<CollectorResult[]> {\n  await googleOAuth.ensureAuthenticated();\n  const calendarApi: calendar_v3.Calendar = google.calendar({\n    version: \"v3\",\n    auth: googleOAuth.getClient(),\n  });\n\n  const calendars = await resolveCalendars(calendarApi, opts.calendars);\n  const lookbackDays = opts.lookbackDays ?? 30;\n  const lookforwardDays = opts.lookforwardDays ?? 90;\n\n  const now = new Date();\n  const timeMin = new Date(\n    now.getTime() - lookbackDays * 24 * 60 * 60 * 1000,\n  ).toISOString();\n  const timeMax = new Date(\n    now.getTime() + lookforwardDays * 24 * 60 * 60 * 1000,\n  ).toISOString();\n\n  console.info(\n    `Fetching events from ${calendars.length} calendar(s): ${calendars.map((c) => c.name).join(\", \")}`,\n  );\n\n  const results: CollectorResult[] = [];\n  let failed = 0;\n\n  for (const cal of calendars) {\n    try {\n      const res = await calendarApi.events.list({\n        calendarId: cal.id,\n        timeMin,\n        timeMax,\n        singleEvents: true,\n        orderBy: \"startTime\",\n        maxResults: 250,\n      });\n\n      const items = res.data.items ?? [];\n      for (const event of items) {\n        const id = event.id ?? \"\";\n        const summary = event.summary ?? \"(no title)\";\n        const startStr = event.start\n          ? formatEventTime(event.start)\n          : \"\";\n        const endStr = event.end ? formatEventTime(event.end) : \"\";\n        const location = event.location ?? \"\";\n        const attendees = (event.attendees ?? [])\n          .map((a) => a.email ?? a.displayName ?? \"\")\n          .filter(Boolean)\n          .join(\", \");\n        const status = event.status ?? \"confirmed\";\n\n        const lines = [\n          `- **Calendar**: ${cal.name}`,\n          startStr ? `- **Time**: ${startStr}${endStr ? ` - ${endStr}` : \"\"}` : \"\",\n          location ? `- **Location**: ${location}` : \"\",\n          attendees ? `- **Attendees**: ${attendees}` : \"\",\n          `- **Status**: ${status}`,\n        ].filter(Boolean);\n\n        results.push({\n          id,\n          title: summary,\n          content: lines.join(\"\\n\") || \"(no details)\",\n        });\n      }\n    } catch (err) {\n      failed++;\n      const msgText = err instanceof Error ? err.message : String(err);\n      console.warn(`Calendar fetch failed for ${cal.name} (${cal.id}): ${msgText}`);\n    }\n  }\n\n  if (failed > 0 && calendars.length > failed) {\n    console.warn(`Calendar fetch failed for ${failed} of ${calendars.length} calendar(s).`);\n  }\n  return results;\n}\n"],"mappings":";;;;;;;;;;;;;AAuBA,SAAS,gBAAgB,OAAiD;CACxE,MAAM,KAAK,MAAM,YAAY,MAAM;CACnC,MAAM,KAAK,MAAM,YAAY;AAC7B,KAAI,CAAC,GAAI,QAAO;AAChB,QAAO,GAAG,GAAG,IAAI,GAAG;;;AAItB,eAAe,iBACb,aACA,aACyC;AACzC,KAAI,YACF,QAAO,YAAY,KAAK,QAAQ;EAAE;EAAI,MAAM;EAAI,EAAE;AAMpD,UAFY,MAAM,YAAY,aAAa,KAAK,EAAE,YAAY,OAAO,CAAC,EACpD,KAAK,SAAS,EAAE,EAE/B,QAAQ,MAAM,EAAE,GAAG,CACnB,KAAK,OAAO;EACX,IAAI,EAAE;EACN,MAAM,EAAE,WAAW,EAAE;EACtB,EAAE;;;;;;AAOP,eAAsB,aACpB,OAA4B,EAAE,EACF;AAC5B,OAAM,YAAY,qBAAqB;CACvC,MAAM,cAAoC,OAAO,SAAS;EACxD,SAAS;EACT,MAAM,YAAY,WAAW;EAC9B,CAAC;CAEF,MAAM,YAAY,MAAM,iBAAiB,aAAa,KAAK,UAAU;CACrE,MAAM,eAAe,KAAK,gBAAgB;CAC1C,MAAM,kBAAkB,KAAK,mBAAmB;CAEhD,MAAM,sBAAM,IAAI,MAAM;CACtB,MAAM,2BAAU,IAAI,KAClB,IAAI,SAAS,GAAG,eAAe,KAAK,KAAK,KAAK,IAC/C,EAAC,aAAa;CACf,MAAM,UAAU,IAAI,KAClB,IAAI,SAAS,GAAG,kBAAkB,KAAK,KAAK,KAAK,IAClD,CAAC,aAAa;AAEf,MACE,wBAAwB,UAAU,OAAO,gBAAgB,UAAU,KAAK,MAAM,EAAE,KAAK,CAAC,KAAK,KAAK,GACjG;CAED,MAAM,UAA6B,EAAE;CACrC,IAAI,SAAS;AAEb,MAAK,MAAM,OAAO,UAChB,KAAI;EAUF,MAAM,SATM,MAAM,YAAY,OAAO,KAAK;GACxC,YAAY,IAAI;GAChB;GACA;GACA,cAAc;GACd,SAAS;GACT,YAAY;GACb,CAAC,EAEgB,KAAK,SAAS,EAAE;AAClC,OAAK,MAAM,SAAS,OAAO;GACzB,MAAM,KAAK,MAAM,MAAM;GACvB,MAAM,UAAU,MAAM,WAAW;GACjC,MAAM,WAAW,MAAM,QACnB,gBAAgB,MAAM,MAAM,GAC5B;GACJ,MAAM,SAAS,MAAM,MAAM,gBAAgB,MAAM,IAAI,GAAG;GACxD,MAAM,WAAW,MAAM,YAAY;GACnC,MAAM,aAAa,MAAM,aAAa,EAAE,EACrC,KAAK,MAAM,EAAE,SAAS,EAAE,eAAe,GAAG,CAC1C,OAAO,QAAQ,CACf,KAAK,KAAK;GACb,MAAM,SAAS,MAAM,UAAU;GAE/B,MAAM,QAAQ;IACZ,mBAAmB,IAAI;IACvB,WAAW,eAAe,WAAW,SAAS,MAAM,WAAW,OAAO;IACtE,WAAW,mBAAmB,aAAa;IAC3C,YAAY,oBAAoB,cAAc;IAC9C,iBAAiB;IAClB,CAAC,OAAO,QAAQ;AAEjB,WAAQ,KAAK;IACX;IACA,OAAO;IACP,SAAS,MAAM,KAAK,KAAK,IAAI;IAC9B,CAAC;;UAEG,KAAK;AACZ;EACA,MAAM,UAAU,eAAe,QAAQ,IAAI,UAAU,OAAO,IAAI;AAChE,OAAa,6BAA6B,IAAI,KAAK,IAAI,IAAI,GAAG,KAAK,UAAU;;AAIjF,KAAI,SAAS,KAAK,UAAU,SAAS,OACnC,MAAa,6BAA6B,OAAO,MAAM,UAAU,OAAO,eAAe;AAEzF,QAAO"}