feat: add user commands, auto-detect browser, optimize performance

- Add user/user-posts/followers/following commands - Add UserProfile model and GraphQL API methods - Add print_user_profile and print_user_table formatters - Auto-detect browser for cookies (Chrome → Edge → Firefox → Brave) - Remove --browser option from all commands - Remove cookie verification (v1.1 endpoints are gone) - Use hardcoded fallback query IDs first (skip slow JS bundle scan) - Update FEATURES from latest twitter-openapi config - Fix user-posts: add required withVoice variable - Add tweet URL links in feed output - Add error handling to all user commands
2026-03-05 00:41:26 +08:00
parent 16752c3115
commit 7238b932ab
10 changed files with 770 additions and 353 deletions
--- a/README.md
+++ b/README.md
@@ -1,6 +1,6 @@
 # Twitter CLI

-从你的 Twitter/X 首页抓取推文，智能筛选高价值内容，AI 自动生成摘要。
+Twitter/X 命令行工具 — 读取 Timeline、管理推文。

 **零 API Key** — 使用浏览器 Cookie 认证，免费访问 Twitter。

@@ -19,38 +19,70 @@ twitter feed

 ## 使用方式

+### 读取
+
 ```bash
-# 完整 pipeline：抓取 50 条 → 筛选 top 20 → AI 总结
+# 抓取首页 timeline
 twitter feed

 # 自定义抓取条数
-twitter feed --count 50
-
-# 只抓取 + 筛选，跳过 AI 总结
-twitter feed --no-summary
-
-# JSON 输出（可重定向到文件）
-twitter feed --json > tweets.json
-
-# 对已有数据做筛选 + 总结
-twitter feed --input tweets.json
+twitter feed --max 50

 # 跳过筛选
 twitter feed --no-filter

-# 指定浏览器
-twitter feed --browser firefox
+# JSON 输出
+twitter feed --json > tweets.json
+
+# 从已有数据加载
+twitter feed --input tweets.json
+

 # 抓取收藏
-twitter bookmarks
-twitter bookmarks --count 30 --json
+twitter favorite
+twitter favorite --max 30 --json
+```
+
+### 用户
+
+```bash
+# 查看用户资料
+twitter user elonmusk
+
+# 列出用户推文
+twitter user-posts elonmusk --max 20
+
+# 查看粉丝
+twitter followers elonmusk --max 30
+
+# 查看关注
+twitter following elonmusk --max 30
+```
+
+### 发推
+
+```bash
+# 发新推文
+twitter post "Hello World"
+
+# 回复推文
+twitter reply <tweet_id> "这是回复内容"
+
+# 引用转推（传 URL 或 ID 均可）
+twitter quote <tweet_url_or_id> "这是引用内容"
+
+# 删除推文（会有确认提示）
+twitter delete <tweet_id>
+
+# 跳过删除确认
+twitter delete <tweet_id> --yes
 ```

 ## Pipeline

 ```
-抓取 (GraphQL API)  →  筛选 (Engagement Score)  →  AI 总结
-      50 条               top 20                  按主题分组
+抓取 (GraphQL API)  →  筛选 (Engagement Score)
+      50 条               top 20
 ```

 ### 筛选算法
@@ -62,10 +94,6 @@ score = 1.0 × likes + 3.0 × retweets + 2.0 × replies
      + 5.0 × bookmarks + 0.5 × log10(views)
 ```

-### AI 总结
-
-支持 **OpenAI-compatible**（doubao / deepseek / openai）和 **Anthropic**（Claude）两种 API 格式。
-
 ## 配置

 编辑 `config.yaml`：
@@ -83,18 +111,11 @@ filter:
    replies: 2.0
    bookmarks: 5.0
    views_log: 0.5
-
-ai:
-  provider: "openai"    # "openai" or "anthropic"
-  api_key: ""           # 或设置环境变量 AI_API_KEY
-  model: "doubao-seed-2.0-code"
-  base_url: "https://ark.cn-beijing.volces.com/api/coding"
-  language: "zh-CN"
 ```

 ### Cookie 配置

-**方式 1：自动提取**（推荐） — 确保 Chrome 已登录 x.com，程序自动通过 `browser-cookie3` 读取。
+**方式 1：自动提取**（推荐） — 确保浏览器已登录 x.com，程序自动通过 `browser-cookie3` 按 Chrome → Edge → Firefox → Brave 顺序尝试读取。

 **方式 2：环境变量** — 设置：

@@ -111,10 +132,9 @@ export TWITTER_CT0=your_ct0
 twitter_cli/
 ├── __init__.py     # 版本信息
 ├── cli.py          # CLI 入口 (click)
-├── client.py       # Twitter GraphQL API Client
+├── client.py       # Twitter GraphQL API Client (GET + POST)
 ├── auth.py         # Cookie 提取 (env / browser-cookie3)
 ├── filter.py       # Engagement scoring + 筛选
-├── summarizer.py   # AI 总结 (OpenAI + Anthropic)
 ├── formatter.py    # Rich 终端输出 + JSON
 ├── config.py       # YAML 配置加载
 └── models.py       # 数据模型 (dataclass)
--- a/config.yaml
+++ b/config.yaml
@@ -13,10 +13,3 @@ filter:
    replies: 2.0
    bookmarks: 5.0
    views_log: 0.5
-
-ai:
-  provider: "openai"
-  api_key: ""
-  model: "doubao-seed-2.0-code"
-  base_url: "https://ark.cn-beijing.volces.com/api/coding"
-  language: "zh-CN"
--- a/pyproject.toml
+++ b/pyproject.toml
@@ -5,7 +5,7 @@ build-backend = "hatchling.build"
 [project]
 name = "twitter-cli"
 version = "0.1.0"
-description = "A CLI for Twitter/X — feed, bookmarks, filtering, AI summary"
+description = "A CLI for Twitter/X — feed, bookmarks, tweet, filtering"
 readme = "README.md"
 license = "Apache-2.0"
 requires-python = ">=3.8"
--- a/twitter_cli/auth.py
+++ b/twitter_cli/auth.py
@@ -8,10 +8,21 @@ Supports:
 from __future__ import annotations

 import json
+import logging
 import os
+import ssl
 import subprocess
 import sys
-from typing import Dict, Optional
+import urllib.request
+from typing import Any, Dict, Optional
+
+logger = logging.getLogger(__name__)
+
+# Public bearer token (same as in client.py)
+_BEARER_TOKEN = (
+    "AAAAAAAAAAAAAAAAAAAAANRILgAAAAAAnNwIzUejRCOuH5E6I8xnZz4puTs"
+    "%3D1Zv7ttfk8LF81IUq16cHjhLTvJu4FA33AGWWjCpTnA"
+)


 def load_from_env() -> Optional[Dict[str, str]]:
@@ -23,9 +34,63 @@ def load_from_env() -> Optional[Dict[str, str]]:
    return None


-def extract_from_browser(browser: str = "chrome") -> Optional[Dict[str, str]]:
+def verify_cookies(auth_token, ct0):
+    # type: (str, str) -> Dict[str, Any]
+    """Verify cookies by calling a Twitter API endpoint.
+
+    Tries multiple endpoints. Only raises on clear auth failures (401/403).
+    For other errors (404, network), returns empty dict (proceed without verification).
+    """
+    # Endpoints to try, in order of preference
+    urls = [
+        "https://api.x.com/1.1/account/verify_credentials.json",
+        "https://x.com/i/api/1.1/account/settings.json",
+    ]
+
+    headers = {
+        "Authorization": "Bearer %s" % _BEARER_TOKEN,
+        "Cookie": "auth_token=%s; ct0=%s" % (auth_token, ct0),
+        "X-Csrf-Token": ct0,
+        "X-Twitter-Active-User": "yes",
+        "X-Twitter-Auth-Type": "OAuth2Session",
+        "User-Agent": (
+            "Mozilla/5.0 (Macintosh; Intel Mac OS X 10_15_7) "
+            "AppleWebKit/537.36 (KHTML, like Gecko) "
+            "Chrome/131.0.0.0 Safari/537.36"
+        ),
+    }
+
+    for url in urls:
+        req = urllib.request.Request(url)
+        for k, v in headers.items():
+            req.add_header(k, v)
+
+        ctx = ssl.create_default_context()
+        try:
+            with urllib.request.urlopen(req, context=ctx, timeout=3) as resp:
+                data = json.loads(resp.read().decode("utf-8"))
+                return {"screen_name": data.get("screen_name", "")}
+        except urllib.error.HTTPError as e:
+            if e.code in (401, 403):
+                raise RuntimeError(
+                    "Cookie expired or invalid (HTTP %d). Please re-login to x.com in your browser." % e.code
+                )
+            # 404 or other — try next endpoint
+            logger.debug("Verification endpoint %s returned HTTP %d, trying next...", url, e.code)
+            continue
+        except Exception as e:
+            logger.debug("Verification endpoint %s failed: %s", url, e)
+            continue
+
+    # All endpoints failed with non-auth errors — proceed without verification
+    logger.info("Cookie verification skipped (no working endpoint), will verify on first API call")
+    return {}
+
+
+def extract_from_browser() -> Optional[Dict[str, str]]:
    """Auto-extract cookies from local browser using browser-cookie3.

+    Tries browsers in order: Chrome -> Edge -> Firefox -> Brave.
    Runs in a subprocess to avoid SQLite database lock issues when the
    browser is running.
    """
@@ -37,25 +102,18 @@ except ImportError:
    print(json.dumps({"error": "browser-cookie3 not installed"}))
    sys.exit(1)

-browser_funcs = {
-    "chrome": browser_cookie3.chrome,
-    "firefox": browser_cookie3.firefox,
-    "edge": browser_cookie3.edge,
-    "brave": browser_cookie3.brave,
-}
-
-browser_name = "%s"
-fn = browser_funcs.get(browser_name)
-if not fn:
-    print(json.dumps({"error": "Unsupported browser: " + browser_name}))
-    sys.exit(1)
+browsers = [
+    ("chrome", browser_cookie3.chrome),
+    ("edge", browser_cookie3.edge),
+    ("firefox", browser_cookie3.firefox),
+    ("brave", browser_cookie3.brave),
+]

+for name, fn in browsers:
    try:
        jar = fn()
-except Exception as e:
-    print(json.dumps({"error": str(e)}))
-    sys.exit(1)
-
+    except Exception:
+        continue
    result = {}
    for cookie in jar:
        domain = cookie.domain or ""
@@ -64,13 +122,14 @@ for cookie in jar:
                result["auth_token"] = cookie.value
            elif cookie.name == "ct0":
                result["ct0"] = cookie.value
-
    if "auth_token" in result and "ct0" in result:
+        result["browser"] = name
        print(json.dumps(result))
-else:
-    print(json.dumps({"error": "Could not find auth_token and ct0 cookies. Make sure you are logged into x.com in " + browser_name + "."}))
+        sys.exit(0)
+
+print(json.dumps({"error": "No Twitter cookies found in any browser. Make sure you are logged into x.com."}))
 sys.exit(1)
-''' % browser
+'''

    try:
        result = subprocess.run(
@@ -97,29 +156,33 @@ else:
        data = json.loads(output)
        if "error" in data:
            return None
+        logger.info("Found cookies in %s", data.get("browser", "unknown"))
        return {"auth_token": data["auth_token"], "ct0": data["ct0"]}
    except (subprocess.TimeoutExpired, json.JSONDecodeError, KeyError, FileNotFoundError):
        return None


-def get_cookies(browser: str = "chrome") -> Dict[str, str]:
-    """Get Twitter cookies. Priority: env vars -> browser extraction.
+def get_cookies() -> Dict[str, str]:
+    """Get Twitter cookies. Priority: env vars -> browser extraction (Chrome/Edge/Firefox/Brave).

-    Returns dict with 'auth_token' and 'ct0' keys.
    Raises RuntimeError if no cookies found.
    """
+    cookies = None  # type: Optional[Dict[str, str]]
+
    # 1. Try environment variables
-    env_cookies = load_from_env()
-    if env_cookies:
-        return env_cookies
+    cookies = load_from_env()
+    if cookies:
+        logger.info("Loaded cookies from environment variables")

-    # 2. Try browser extraction
-    browser_cookies = extract_from_browser(browser)
-    if browser_cookies:
-        return browser_cookies
+    # 2. Try browser extraction (auto-detect)
+    if not cookies:
+        cookies = extract_from_browser()

+    if not cookies:
        raise RuntimeError(
            "No Twitter cookies found.\n"
            "Option 1: Set TWITTER_AUTH_TOKEN and TWITTER_CT0 environment variables\n"
-        "Option 2: Make sure you are logged into x.com in your browser"
+            "Option 2: Make sure you are logged into x.com in your browser (Chrome/Edge/Firefox/Brave)"
        )
+
+    return cookies
--- a/twitter_cli/cli.py
+++ b/twitter_cli/cli.py
@@ -1,16 +1,18 @@
 """CLI entry point for twitter-cli.

 Usage:
-    twitter feed                      # full pipeline: fetch → filter → AI summarize
-    twitter feed --count 50           # custom fetch count
-    twitter feed --no-summary         # skip AI summary
+    twitter feed                      # fetch home timeline → filter
+    twitter feed --max 50             # custom fetch count
    twitter feed --no-filter          # skip filtering
    twitter feed --json               # JSON output
-    twitter feed --browser firefox    # specify browser for cookie extraction
-    twitter bookmarks                 # fetch bookmarks
-    twitter bookmarks --count 30
-    twitter feed --input tweets.json  # summarize existing data
+    twitter favorite                  # fetch bookmarks
+    twitter favorite --max 30
+    twitter feed --input tweets.json  # load existing data
    twitter feed --output out.json    # save filtered tweets
+    twitter post "Hello"              # post a tweet
+    twitter reply ID "text"           # reply to a tweet
+    twitter quote ID "text"           # quote a tweet
+    twitter delete ID                 # delete a tweet
 """

 from __future__ import annotations
@@ -33,10 +35,12 @@ from .filter import filter_tweets
 from .formatter import (
    print_filter_stats,
    print_tweet_table,
+    print_user_profile,
+    print_user_table,
    tweets_to_json,
 )
 from .models import Author, Metrics, Tweet, TweetMedia
-from .summarizer import summarize
+

 console = Console()

@@ -132,16 +136,14 @@ def cli(verbose):
 # ===== Feed =====

@cli.command()
-@click.option("--count", "-n", type=int, default=None, help="Number of tweets to fetch.")
+@click.option("--max", "-n", "max_count", type=int, default=None, help="Max number of tweets to fetch.")
@click.option("--json", "as_json", is_flag=True, help="Output as JSON.")
-@click.option("--browser", "-b", default="chrome", help="Browser to extract cookies from.")
@click.option("--input", "-i", "input_file", type=str, default=None, help="Load tweets from JSON file.")
@click.option("--output", "-o", "output_file", type=str, default=None, help="Save filtered tweets to JSON file.")
@click.option("--no-filter", is_flag=True, help="Skip filtering.")
-@click.option("--no-summary", is_flag=True, help="Skip AI summary.")
-def feed(count, as_json, browser, input_file, output_file, no_filter, no_summary):
-    # type: (int, bool, str, str, str, bool, bool) -> None
-    """Fetch home timeline — full pipeline: fetch → filter → AI summarize."""
+def feed(max_count, as_json, input_file, output_file, no_filter):
+    # type: (int, bool, str, str, bool) -> None
+    """Fetch home timeline with filtering."""
    config = load_config()

    # Step 1: Get tweets
@@ -150,10 +152,10 @@ def feed(count, as_json, browser, input_file, output_file, no_filter, no_summary
        tweets = _load_tweets_from_json(input_file)
        console.print("   Loaded %d tweets" % len(tweets))
    else:
-        fetch_count = count or config.get("fetch", {}).get("count", 50)
+        fetch_count = max_count or config.get("fetch", {}).get("count", 50)
        console.print("\n🔐 Getting Twitter cookies...")
        try:
-            cookies = get_cookies(browser)
+            cookies = get_cookies()
        except RuntimeError as e:
            console.print("[red]❌ %s[/red]" % e)
            sys.exit(1)
@@ -188,58 +190,33 @@ def feed(count, as_json, browser, input_file, output_file, no_filter, no_summary
    print_tweet_table(filtered, console)
    console.print()

-    # Step 3: AI Summary
-    if no_summary:
-        return

-    ai_config = config.get("ai", {})
-    if not ai_config.get("api_key"):
-        console.print(
-            "[yellow]⚠️  AI summary skipped: no API key configured.[/yellow]\n"
-            "   Set ai.api_key in config.yaml or export AI_API_KEY=your_key"
-        )
-        return
-
-    try:
-        console.print("🤖 Calling AI (%s/%s)..." % (ai_config.get("provider", "openai"), ai_config.get("model", "")))
-        summary = summarize(filtered, ai_config)
-        console.print("\n" + "═" * 50)
-        console.print("📝 AI Summary")
-        console.print("═" * 50 + "\n")
-        console.print(summary)
-        console.print()
-    except Exception as e:
-        console.print("[red]❌ AI summary failed: %s[/red]" % e)
-
-
-# ===== Bookmarks =====
+# ===== Favorite =====

@cli.command()
-@click.option("--count", "-n", type=int, default=None, help="Number of tweets to fetch.")
+@click.option("--max", "-n", "max_count", type=int, default=None, help="Max number of tweets to fetch.")
@click.option("--json", "as_json", is_flag=True, help="Output as JSON.")
-@click.option("--browser", "-b", default="chrome", help="Browser to extract cookies from.")
@click.option("--output", "-o", "output_file", type=str, default=None, help="Save tweets to JSON file.")
@click.option("--no-filter", is_flag=True, help="Skip filtering.")
-@click.option("--no-summary", is_flag=True, help="Skip AI summary.")
-def bookmarks(count, as_json, browser, output_file, no_filter, no_summary):
-    # type: (int, bool, str, str, bool, bool) -> None
-    """Fetch bookmarked tweets."""
+def favorite(max_count, as_json, output_file, no_filter):
+    # type: (int, bool, str, bool) -> None
+    """Fetch bookmarked (favorite) tweets."""
    config = load_config()
-    fetch_count = count or 50
+    fetch_count = max_count or 50

    console.print("\n🔐 Getting Twitter cookies...")
    try:
-        cookies = get_cookies(browser)
+        cookies = get_cookies()
    except RuntimeError as e:
        console.print("[red]❌ %s[/red]" % e)
        sys.exit(1)

    client = TwitterClient(cookies["auth_token"], cookies["ct0"])
-    console.print("🔖 Fetching bookmarks (%d tweets)...\n" % fetch_count)
+    console.print("🔖 Fetching favorites (%d tweets)...\n" % fetch_count)
    start = time.time()
    tweets = client.fetch_bookmarks(fetch_count)
    elapsed = time.time() - start
-    console.print("✅ Fetched %d bookmarks in %.1fs\n" % (len(tweets), elapsed))
+    console.print("✅ Fetched %d favorites in %.1fs\n" % (len(tweets), elapsed))

    # Filter
    if no_filter:
@@ -261,29 +238,204 @@ def bookmarks(count, as_json, browser, output_file, no_filter, no_summary):
        click.echo(tweets_to_json(filtered))
        return

-    print_tweet_table(filtered, console, title="🔖 Bookmarks — %d tweets" % len(filtered))
+    print_tweet_table(filtered, console, title="🔖 Favorites — %d tweets" % len(filtered))
    console.print()

-    # AI Summary
-    if no_summary:
-        return

-    ai_config = config.get("ai", {})
-    if not ai_config.get("api_key"):
-        console.print(
-            "[yellow]⚠️  AI summary skipped: no API key configured.[/yellow]"
-        )
-        return
+# ===== User =====

+@cli.command()
+@click.argument("screen_name")
+def user(screen_name):
+    # type: (str,) -> None
+    """View a user's profile. SCREEN_NAME is the @handle (without @)."""
+    screen_name = screen_name.lstrip("@")
+    client = _get_client()
+    console.print("👤 Fetching user @%s..." % screen_name)
    try:
-        console.print("🤖 Calling AI...")
-        summary = summarize(filtered, ai_config)
-        console.print("\n" + "═" * 50)
-        console.print("📝 AI Summary")
-        console.print("═" * 50 + "\n")
-        console.print(summary)
-    except Exception as e:
-        console.print("[red]❌ AI summary failed: %s[/red]" % e)
+        profile = client.fetch_user(screen_name)
+        console.print()
+        print_user_profile(profile, console)
+    except RuntimeError as e:
+        console.print("[red]❌ %s[/red]" % e)
+        sys.exit(1)
+
+
+@cli.command("user-posts")
+@click.argument("screen_name")
+@click.option("--max", "-n", "max_count", type=int, default=20, help="Max number of tweets to fetch.")
+@click.option("--json", "as_json", is_flag=True, help="Output as JSON.")
+def user_posts(screen_name, max_count, as_json):
+    # type: (str, int, bool) -> None
+    """List a user's tweets. SCREEN_NAME is the @handle (without @)."""
+    screen_name = screen_name.lstrip("@")
+    client = _get_client()
+    console.print("👤 Fetching @%s's profile..." % screen_name)
+    try:
+        profile = client.fetch_user(screen_name)
+    except RuntimeError as e:
+        console.print("[red]❌ %s[/red]" % e)
+        sys.exit(1)
+
+    console.print("📝 Fetching tweets (%d)...\n" % max_count)
+    start = time.time()
+    try:
+        tweets = client.fetch_user_tweets(profile.id, max_count)
+    except RuntimeError as e:
+        console.print("[red]❌ %s[/red]" % e)
+        sys.exit(1)
+    elapsed = time.time() - start
+    console.print("✅ Fetched %d tweets in %.1fs\n" % (len(tweets), elapsed))
+
+    if as_json:
+        click.echo(tweets_to_json(tweets))
+        return
+
+    print_tweet_table(tweets, console, title="📝 @%s — %d tweets" % (screen_name, len(tweets)))
+    console.print()
+
+
+@cli.command()
+@click.argument("screen_name")
+@click.option("--max", "-n", "max_count", type=int, default=20, help="Max number of users to show.")
+def followers(screen_name, max_count):
+    # type: (str, int) -> None
+    """List a user's followers. SCREEN_NAME is the @handle (without @)."""
+    screen_name = screen_name.lstrip("@")
+    client = _get_client()
+    console.print("👤 Fetching @%s's profile..." % screen_name)
+    try:
+        profile = client.fetch_user(screen_name)
+    except RuntimeError as e:
+        console.print("[red]❌ %s[/red]" % e)
+        sys.exit(1)
+
+    console.print("👥 Fetching followers...\n")
+    try:
+        users = client.fetch_followers(profile.id, max_count)
+    except RuntimeError as e:
+        console.print("[red]❌ %s[/red]" % e)
+        sys.exit(1)
+    print_user_table(users, console, title="👥 @%s's followers — %d" % (screen_name, len(users)))
+    console.print()
+
+
+@cli.command()
+@click.argument("screen_name")
+@click.option("--max", "-n", "max_count", type=int, default=20, help="Max number of users to show.")
+def following(screen_name, max_count):
+    # type: (str, int) -> None
+    """List users that someone follows. SCREEN_NAME is the @handle (without @)."""
+    screen_name = screen_name.lstrip("@")
+    client = _get_client()
+    console.print("👤 Fetching @%s's profile..." % screen_name)
+    try:
+        profile = client.fetch_user(screen_name)
+    except RuntimeError as e:
+        console.print("[red]❌ %s[/red]" % e)
+        sys.exit(1)
+
+    console.print("👥 Fetching following...\n")
+    try:
+        users = client.fetch_following(profile.id, max_count)
+    except RuntimeError as e:
+        console.print("[red]❌ %s[/red]" % e)
+        sys.exit(1)
+    print_user_table(users, console, title="👥 @%s follows — %d" % (screen_name, len(users)))
+    console.print()
+
+
+# ===== Post / Reply / Quote / Delete =====
+
+def _get_client():
+    # type: () -> TwitterClient
+    """Helper to authenticate and create a TwitterClient."""
+    console.print("\n🔐 Getting Twitter cookies...")
+    try:
+        cookies = get_cookies()
+    except RuntimeError as e:
+        console.print("[red]❌ %s[/red]" % e)
+        sys.exit(1)
+    return TwitterClient(cookies["auth_token"], cookies["ct0"])
+
+
+@cli.command()
+@click.argument("text")
+def post(text):
+    # type: (str,) -> None
+    """Post a new tweet."""
+    client = _get_client()
+    console.print("✏️  Posting tweet...")
+    try:
+        result = client.create_tweet(text)
+        tweet_id = result["tweet_id"]
+        console.print("\n[green]✅ Tweet posted![/green]")
+        console.print("   ID: %s" % tweet_id)
+        console.print("   URL: https://x.com/i/status/%s" % tweet_id)
+        console.print('   Text: "%s"' % result["text"][:100])
+    except RuntimeError as e:
+        console.print("[red]❌ %s[/red]" % e)
+        sys.exit(1)
+
+
+@cli.command()
+@click.argument("tweet_id")
+@click.argument("text")
+def reply(tweet_id, text):
+    # type: (str, str) -> None
+    """Reply to a tweet."""
+    client = _get_client()
+    console.print("💬 Replying to %s..." % tweet_id)
+    try:
+        result = client.create_tweet(text, reply_to=tweet_id)
+        new_id = result["tweet_id"]
+        console.print("\n[green]✅ Reply posted![/green]")
+        console.print("   ID: %s" % new_id)
+        console.print("   URL: https://x.com/i/status/%s" % new_id)
+        console.print('   Text: "%s"' % result["text"][:100])
+    except RuntimeError as e:
+        console.print("[red]❌ %s[/red]" % e)
+        sys.exit(1)
+
+
+@cli.command()
+@click.argument("tweet_url")
+@click.argument("text")
+def quote(tweet_url, text):
+    # type: (str, str) -> None
+    """Quote a tweet. TWEET_URL can be a full URL or tweet ID."""
+    # If user passes just an ID, convert to URL
+    if not tweet_url.startswith("http"):
+        tweet_url = "https://x.com/i/status/%s" % tweet_url
+    client = _get_client()
+    console.print("🔄 Quoting %s..." % tweet_url)
+    try:
+        result = client.create_tweet(text, quote_tweet_url=tweet_url)
+        new_id = result["tweet_id"]
+        console.print("\n[green]✅ Quote tweet posted![/green]")
+        console.print("   ID: %s" % new_id)
+        console.print("   URL: https://x.com/i/status/%s" % new_id)
+        console.print('   Text: "%s"' % result["text"][:100])
+    except RuntimeError as e:
+        console.print("[red]❌ %s[/red]" % e)
+        sys.exit(1)
+
+
+@cli.command()
+@click.argument("tweet_id")
+@click.confirmation_option(prompt="Are you sure you want to delete this tweet?")
+def delete(tweet_id):
+    # type: (str,) -> None
+    """Delete a tweet by ID."""
+    client = _get_client()
+    console.print("🗑️  Deleting tweet %s..." % tweet_id)
+    try:
+        client.delete_tweet(tweet_id)
+        console.print("\n[green]✅ Tweet deleted![/green]")
+        console.print("   ID: %s" % tweet_id)
+    except RuntimeError as e:
+        console.print("[red]❌ %s[/red]" % e)
+        sys.exit(1)


 if __name__ == "__main__":
--- a/twitter_cli/client.py
+++ b/twitter_cli/client.py
@@ -15,7 +15,7 @@ import ssl
 import urllib.request
 from typing import Any, Callable, Dict, List, Optional, Tuple

-from .models import Author, Metrics, Tweet, TweetMedia
+from .models import Author, Metrics, Tweet, TweetMedia, UserProfile

 logger = logging.getLogger(__name__)

@@ -27,8 +27,14 @@ BEARER_TOKEN = (

 # Last-resort fallback query IDs
 FALLBACK_QUERY_IDS = {
-    "HomeTimeline": "HJFjzBgCs16TqxewQOeLNg",
+    "HomeTimeline": "c-CzHF1LboFilMpsx4ZCrQ",
    "Bookmarks": "VFdMm9iVZxlU6hD86gfW_A",
+    "CreateTweet": "oB-5XsHNAbjvARJEc8CZFw",
+    "DeleteTweet": "VaenaVgh5q5ih7kvyVjgtg",
+    "UserByScreenName": "1VOOyvKkiI3FMmkeDNxM9A",
+    "UserTweets": "E3opETHurmVJflFsUBVuUQ",
+    "Followers": "IOh4aS6UdGWGJUYTqliQ7Q",
+    "Following": "zx6e-TLzRkeDO_a7p4b3JQ",
 }

 # Community-maintained API definition (auto-updated daily)
@@ -39,14 +45,20 @@ TWITTER_OPENAPI_URL = (

 # Default features flags required by the GraphQL endpoint
 FEATURES = {
+    "rweb_video_screen_enabled": False,
+    "profile_label_improvements_pcf_label_in_post_enabled": True,
    "rweb_tipjar_consumption_enabled": True,
-    "responsive_web_graphql_exclude_directive_enabled": True,
    "verified_phone_label_enabled": False,
    "creator_subscriptions_tweet_preview_api_enabled": True,
    "responsive_web_graphql_timeline_navigation_enabled": True,
    "responsive_web_graphql_skip_user_profile_image_extensions_enabled": False,
+    "premium_content_api_read_enabled": False,
    "communities_web_enable_tweet_community_results_fetch": True,
    "c9s_tweet_anatomy_moderator_badge_enabled": True,
+    "responsive_web_grok_analyze_button_fetch_trends_enabled": False,
+    "responsive_web_grok_analyze_post_followups_enabled": True,
+    "responsive_web_jetfuel_frame": False,
+    "responsive_web_grok_share_attachment_enabled": True,
    "articles_preview_enabled": True,
    "responsive_web_edit_tweet_api_enabled": True,
    "graphql_is_translatable_rweb_tweet_is_translatable_enabled": True,
@@ -54,13 +66,15 @@ FEATURES = {
    "longform_notetweets_consumption_enabled": True,
    "responsive_web_twitter_article_tweet_consumption_enabled": True,
    "tweet_awards_web_tipping_enabled": False,
+    "responsive_web_grok_show_grok_translated_post": False,
+    "responsive_web_grok_analysis_button_from_backend": False,
    "creator_subscriptions_quote_tweet_preview_enabled": False,
    "freedom_of_speech_not_reach_fetch_enabled": True,
    "standardized_nudges_misinfo": True,
    "tweet_with_visibility_results_prefer_gql_limited_actions_policy_enabled": True,
-    "rweb_video_timestamps_enabled": True,
    "longform_notetweets_rich_text_read_enabled": True,
    "longform_notetweets_inline_media_enabled": True,
+    "responsive_web_grok_image_annotation_enabled": True,
    "responsive_web_enhance_cards_enabled": False,
 }

@@ -152,17 +166,18 @@ def _fetch_from_github(operation_name):

 def _resolve_query_id(operation_name):
    # type: (str) -> str
-    """Resolve queryId using three-tier strategy: bundle scan -> GitHub -> fallback."""
+    """Resolve queryId using three-tier strategy: fallback -> GitHub -> bundle scan."""
    if operation_name in _cached_query_ids:
        return _cached_query_ids[operation_name]

-    logger.info("Auto-detecting %s queryId...", operation_name)
+    # Tier 1: Hardcoded fallback (instant, no network)
+    fallback = FALLBACK_QUERY_IDS.get(operation_name)
+    if fallback:
+        logger.debug("Using fallback queryId for %s: %s", operation_name, fallback)
+        _cached_query_ids[operation_name] = fallback
+        return fallback

-    # Tier 1: JS bundle scan
-    _scan_bundles()
-    if operation_name in _cached_query_ids:
-        logger.info("Found %s queryId: %s", operation_name, _cached_query_ids[operation_name])
-        return _cached_query_ids[operation_name]
+    logger.info("Auto-detecting %s queryId (no fallback available)...", operation_name)

    # Tier 2: GitHub
    github_id = _fetch_from_github(operation_name)
@@ -170,12 +185,11 @@ def _resolve_query_id(operation_name):
        _cached_query_ids[operation_name] = github_id
        return github_id

-    # Tier 3: Hardcoded fallback
-    fallback = FALLBACK_QUERY_IDS.get(operation_name)
-    if fallback:
-        logger.info("Using hardcoded fallback queryId for %s: %s", operation_name, fallback)
-        _cached_query_ids[operation_name] = fallback
-        return fallback
+    # Tier 3: JS bundle scan
+    _scan_bundles()
+    if operation_name in _cached_query_ids:
+        logger.info("Found %s queryId: %s", operation_name, _cached_query_ids[operation_name])
+        return _cached_query_ids[operation_name]

    raise RuntimeError(
        'Cannot resolve queryId for "%s" — all detection methods failed' % operation_name
@@ -291,6 +305,260 @@ class TwitterClient:
            body = e.read().decode("utf-8", errors="replace")
            raise RuntimeError("Twitter API error %d: %s" % (e.code, body[:500]))

+    def _api_post(self, url, payload):
+        # type: (str, Dict[str, Any]) -> Any
+        """Make authenticated POST request to Twitter API."""
+        headers = self._build_headers()
+        data = json.dumps(payload).encode("utf-8")
+        req = urllib.request.Request(url, data=data, method="POST")
+        for k, v in headers.items():
+            req.add_header(k, v)
+
+        ctx = _create_ssl_context()
+        try:
+            with urllib.request.urlopen(req, context=ctx, timeout=30) as resp:
+                body = resp.read().decode("utf-8")
+                return json.loads(body)
+        except urllib.error.HTTPError as e:
+            body = e.read().decode("utf-8", errors="replace")
+            raise RuntimeError("Twitter API error %d: %s" % (e.code, body[:500]))
+
+    def create_tweet(self, text, reply_to=None, quote_tweet_url=None):
+        # type: (str, Optional[str], Optional[str]) -> Dict[str, Any]
+        """Create a tweet, reply, or quote tweet.
+
+        Args:
+            text: Tweet text content.
+            reply_to: Tweet ID to reply to (optional).
+            quote_tweet_url: URL of tweet to quote (optional).
+
+        Returns:
+            Dict with tweet_id and text of the created tweet.
+        """
+        query_id = _resolve_query_id("CreateTweet")
+        url = "https://x.com/i/api/graphql/%s/CreateTweet" % query_id
+
+        variables = {
+            "tweet_text": text,
+            "dark_request": False,
+            "media": {"media_entities": [], "possibly_sensitive": False},
+            "semantic_annotation_ids": [],
+        }  # type: Dict[str, Any]
+
+        if reply_to:
+            variables["reply"] = {
+                "in_reply_to_tweet_id": reply_to,
+                "exclude_reply_user_ids": [],
+            }
+
+        if quote_tweet_url:
+            variables["attachment_url"] = quote_tweet_url
+
+        features = {
+            "communities_web_enable_tweet_community_results_fetch": True,
+            "c9s_tweet_anatomy_moderator_badge_enabled": True,
+            "responsive_web_edit_tweet_api_enabled": True,
+            "graphql_is_translatable_rweb_tweet_is_translatable_enabled": True,
+            "view_counts_everywhere_api_enabled": True,
+            "longform_notetweets_consumption_enabled": True,
+            "responsive_web_twitter_article_tweet_consumption_enabled": True,
+            "tweet_awards_web_tipping_enabled": False,
+            "creator_subscriptions_quote_tweet_preview_enabled": False,
+            "longform_notetweets_rich_text_read_enabled": True,
+            "longform_notetweets_inline_media_enabled": True,
+            "articles_preview_enabled": True,
+            "rweb_video_timestamps_enabled": True,
+            "rweb_tipjar_consumption_enabled": True,
+            "responsive_web_graphql_exclude_directive_enabled": True,
+            "verified_phone_label_enabled": False,
+            "freedom_of_speech_not_reach_fetch_enabled": True,
+            "standardized_nudges_misinfo": True,
+            "tweet_with_visibility_results_prefer_gql_limited_actions_policy_enabled": True,
+            "responsive_web_graphql_skip_user_profile_image_extensions_enabled": False,
+            "responsive_web_graphql_timeline_navigation_enabled": True,
+            "responsive_web_enhance_cards_enabled": False,
+        }
+
+        payload = {
+            "variables": variables,
+            "features": features,
+            "queryId": query_id,
+        }
+
+        data = self._api_post(url, payload)
+
+        # Parse response
+        result = _deep_get(data, "data", "create_tweet", "tweet_results", "result")
+        if not result:
+            errors = data.get("errors", [])
+            if errors:
+                raise RuntimeError("CreateTweet failed: %s" % errors[0].get("message", str(errors)))
+            raise RuntimeError("CreateTweet failed: unexpected response")
+
+        tweet_id = result.get("rest_id", "")
+        tweet_text = _deep_get(result, "legacy", "full_text") or text
+        return {"tweet_id": tweet_id, "text": tweet_text}
+
+    def delete_tweet(self, tweet_id):
+        # type: (str) -> bool
+        """Delete a tweet by ID.
+
+        Returns:
+            True if deletion was successful.
+        """
+        query_id = _resolve_query_id("DeleteTweet")
+        url = "https://x.com/i/api/graphql/%s/DeleteTweet" % query_id
+
+        payload = {
+            "variables": {"tweet_id": tweet_id, "dark_request": False},
+            "queryId": query_id,
+        }
+
+        data = self._api_post(url, payload)
+
+        # Check response
+        result = _deep_get(data, "data", "delete_tweet", "tweet_results")
+        if result is not None:
+            return True
+
+        errors = data.get("errors", [])
+        if errors:
+            raise RuntimeError("DeleteTweet failed: %s" % errors[0].get("message", str(errors)))
+        # Some successful deletions return empty result
+        return True
+
+    def fetch_user(self, screen_name):
+        # type: (str) -> UserProfile
+        """Fetch user profile by screen name."""
+        query_id = _resolve_query_id("UserByScreenName")
+        variables = {
+            "screen_name": screen_name,
+            "withSafetyModeUserFields": True,
+        }
+        features = {
+            "hidden_profile_subscriptions_enabled": True,
+            "rweb_tipjar_consumption_enabled": True,
+            "responsive_web_graphql_exclude_directive_enabled": True,
+            "verified_phone_label_enabled": False,
+            "subscriptions_verification_info_is_identity_verified_enabled": True,
+            "subscriptions_verification_info_verified_since_enabled": True,
+            "highlights_tweets_tab_ui_enabled": True,
+            "responsive_web_twitter_article_notes_tab_enabled": True,
+            "subscriptions_feature_can_gift_premium": True,
+            "creator_subscriptions_tweet_preview_api_enabled": True,
+            "responsive_web_graphql_skip_user_profile_image_extensions_enabled": False,
+            "responsive_web_graphql_timeline_navigation_enabled": True,
+        }
+
+        url = "https://x.com/i/api/graphql/%s/UserByScreenName?" % query_id
+        url += "variables=%s&features=%s" % (
+            urllib.request.quote(json.dumps(variables)),
+            urllib.request.quote(json.dumps(features)),
+        )
+
+        data = self._api_get(url)
+        result = _deep_get(data, "data", "user", "result")
+        if not result:
+            raise RuntimeError("User @%s not found" % screen_name)
+
+        legacy = result.get("legacy", {})
+        core = result.get("core", {})
+        return UserProfile(
+            id=result.get("rest_id", ""),
+            name=core.get("name") or legacy.get("name", ""),
+            screen_name=core.get("screen_name") or legacy.get("screen_name", screen_name),
+            bio=legacy.get("description", ""),
+            location=legacy.get("location", ""),
+            url=(
+                legacy.get("entities", {}).get("url", {}).get("urls", [{}])[0].get("expanded_url", "")
+                if legacy.get("entities", {}).get("url")
+                else ""
+            ),
+            followers_count=legacy.get("followers_count", 0),
+            following_count=legacy.get("friends_count", 0),
+            tweets_count=legacy.get("statuses_count", 0),
+            likes_count=legacy.get("favourites_count", 0),
+            verified=bool(result.get("is_blue_verified") or legacy.get("verified", False)),
+            profile_image_url=legacy.get("profile_image_url_https", ""),
+            created_at=legacy.get("created_at", ""),
+        )
+
+    def fetch_user_tweets(self, user_id, count=20):
+        # type: (str, int) -> List[Tweet]
+        """Fetch tweets posted by a user."""
+        query_id = _resolve_query_id("UserTweets")
+        return self._fetch_timeline(
+            query_id,
+            "UserTweets",
+            count,
+            lambda data: _deep_get(data, "data", "user", "result", "timeline_v2", "timeline", "instructions"),
+            extra_variables={
+                "userId": user_id,
+                "withQuickPromoteEligibilityTweetFields": True,
+                "withVoice": True,
+                "withV2Timeline": True,
+            },
+        )
+
+    def fetch_followers(self, user_id, count=20):
+        # type: (str, int) -> List[UserProfile]
+        """Fetch user's followers."""
+        query_id = _resolve_query_id("Followers")
+        return self._fetch_user_list(query_id, "Followers", user_id, count)
+
+    def fetch_following(self, user_id, count=20):
+        # type: (str, int) -> List[UserProfile]
+        """Fetch users that this user follows."""
+        query_id = _resolve_query_id("Following")
+        return self._fetch_user_list(query_id, "Following", user_id, count)
+
+    def _fetch_user_list(self, query_id, operation_name, user_id, count):
+        # type: (str, str, str, int) -> List[UserProfile]
+        """Generic user list fetcher (followers/following)."""
+        variables = {
+            "userId": user_id,
+            "count": min(count, 50),
+            "includePromotedContent": False,
+        }  # type: Dict[str, Any]
+
+        url = "https://x.com/i/api/graphql/%s/%s?" % (query_id, operation_name)
+        url += "variables=%s&features=%s" % (
+            urllib.request.quote(json.dumps(variables)),
+            urllib.request.quote(json.dumps(FEATURES)),
+        )
+
+        data = self._api_get(url)
+        users = []  # type: List[UserProfile]
+
+        instructions = _deep_get(data, "data", "user", "result", "timeline", "timeline", "instructions")
+        if not isinstance(instructions, list):
+            return users
+
+        for instruction in instructions:
+            entries = instruction.get("entries", [])
+            for entry in entries:
+                content = entry.get("content", {})
+                item_content = content.get("itemContent", {})
+                user_results = item_content.get("user_results", {}).get("result")
+                if not user_results:
+                    continue
+                legacy = user_results.get("legacy", {})
+                core = user_results.get("core", {})
+                if not legacy:
+                    continue
+                users.append(UserProfile(
+                    id=user_results.get("rest_id", ""),
+                    name=core.get("name") or legacy.get("name", ""),
+                    screen_name=core.get("screen_name") or legacy.get("screen_name", ""),
+                    bio=legacy.get("description", ""),
+                    followers_count=legacy.get("followers_count", 0),
+                    following_count=legacy.get("friends_count", 0),
+                    verified=bool(user_results.get("is_blue_verified") or legacy.get("verified", False)),
+                    profile_image_url=legacy.get("profile_image_url_https", ""),
+                ))
+
+        return users[:count]
+
    def _parse_timeline_response(self, data, get_instructions):
        # type: (Any, Callable) -> Tuple[List[Tweet], Optional[str]]
        """Parse timeline GraphQL response into tweets + next cursor."""
--- a/twitter_cli/config.py
+++ b/twitter_cli/config.py
@@ -5,7 +5,6 @@ Uses a simple built-in YAML parser to avoid adding PyYAML as a dependency.

 from __future__ import annotations

-import os
 import re
 from pathlib import Path
 from typing import Any, Dict, List, Union
@@ -29,13 +28,6 @@ DEFAULT_CONFIG = {
            "views_log": 0.5,
        },
    },
-    "ai": {
-        "provider": "openai",
-        "api_key": "",
-        "model": "doubao-seed-2.0-code",
-        "base_url": "https://ark.cn-beijing.volces.com/api/coding",
-        "language": "zh-CN",
-    },
 }  # type: Dict[str, Any]


@@ -161,15 +153,9 @@ def load_config(config_path=None):
    # Ensure nested dicts exist
    config.setdefault("fetch", DEFAULT_CONFIG["fetch"])
    config.setdefault("filter", DEFAULT_CONFIG["filter"])
-    config.setdefault("ai", DEFAULT_CONFIG["ai"])

    # Deep-copy filter weights if needed
    if "filter" in config and "weights" not in config["filter"]:
        config["filter"]["weights"] = dict(DEFAULT_CONFIG["filter"]["weights"])

-    # AI API key fallback to env var
-    ai = config.get("ai", {})
-    if not ai.get("api_key"):
-        ai["api_key"] = os.environ.get("AI_API_KEY", "")
-
    return config
--- a/twitter_cli/formatter.py
+++ b/twitter_cli/formatter.py
@@ -10,7 +10,7 @@ from rich.panel import Panel
 from rich.table import Table
 from rich.text import Text

-from .models import Tweet
+from .models import Tweet, UserProfile


 def format_number(n):
@@ -69,6 +69,9 @@ def print_tweet_table(tweets, console=None, title=None):
            qt_text = qt.text.replace("\n", " ")[:60]
            text += "\n┌ @%s: %s" % (qt.author.screen_name, qt_text)

+        # Tweet link
+        text += "\n🔗 x.com/%s/status/%s" % (tweet.author.screen_name, tweet.id)
+
        # Stats
        stats = (
            "❤️ %s  🔄 %s\n💬 %s  👁️ %s"
@@ -205,3 +208,82 @@ def tweets_to_json(tweets):
            }
        result.append(d)
    return json.dumps(result, ensure_ascii=False, indent=2)
+
+
+def print_user_profile(user, console=None):
+    # type: (UserProfile, Optional[Console]) -> None
+    """Print user profile as a rich panel."""
+    if console is None:
+        console = Console()
+
+    verified = " ✓" if user.verified else ""
+    header = "@%s%s (%s)" % (user.screen_name, verified, user.name)
+
+    lines = []
+    if user.bio:
+        lines.append(user.bio)
+        lines.append("")
+
+    if user.location:
+        lines.append("📍 %s" % user.location)
+    if user.url:
+        lines.append("🔗 %s" % user.url)
+    if user.location or user.url:
+        lines.append("")
+
+    lines.append(
+        "👥 %s followers · %s following · %s tweets · %s likes"
+        % (
+            format_number(user.followers_count),
+            format_number(user.following_count),
+            format_number(user.tweets_count),
+            format_number(user.likes_count),
+        )
+    )
+
+    if user.created_at:
+        lines.append("📅 Joined %s" % user.created_at)
+    lines.append("🔗 x.com/%s" % user.screen_name)
+
+    console.print(Panel(
+        "\n".join(lines),
+        title=header,
+        border_style="cyan",
+        expand=True,
+    ))
+
+
+def print_user_table(users, console=None, title=None):
+    # type: (List[UserProfile], Optional[Console], Optional[str]) -> None
+    """Print a list of users as a rich table."""
+    if console is None:
+        console = Console()
+
+    if not title:
+        title = "👥 Users — %d" % len(users)
+
+    table = Table(title=title, show_lines=True, expand=True)
+    table.add_column("#", style="dim", width=3, justify="right")
+    table.add_column("User", style="cyan", width=20, no_wrap=True)
+    table.add_column("Bio", ratio=3)
+    table.add_column("Stats", style="green", width=22, no_wrap=True)
+
+    for i, user in enumerate(users):
+        verified = " ✓" if user.verified else ""
+        user_text = "@%s%s\n%s" % (user.screen_name, verified, user.name)
+
+        bio = (user.bio or "").replace("\n", " ").strip()
+        if len(bio) > 100:
+            bio = bio[:97] + "..."
+
+        stats = (
+            "👥 %s followers\n📝 %s following"
+            % (
+                format_number(user.followers_count),
+                format_number(user.following_count),
+            )
+        )
+
+        table.add_row(str(i + 1), user_text, bio, stats)
+
+    console.print(table)
--- a/twitter_cli/models.py
+++ b/twitter_cli/models.py
@@ -50,3 +50,20 @@ class Tweet:
    retweeted_by: Optional[str] = None
    quoted_tweet: Optional[Tweet] = None
    score: float = 0.0
+
+
+@dataclass
+class UserProfile:
+    id: str
+    name: str
+    screen_name: str
+    bio: str = ""
+    location: str = ""
+    url: str = ""
+    followers_count: int = 0
+    following_count: int = 0
+    tweets_count: int = 0
+    likes_count: int = 0
+    verified: bool = False
+    profile_image_url: str = ""
+    created_at: str = ""
--- a/twitter_cli/summarizer.py
+++ b/twitter_cli/summarizer.py
@@ -1,164 +0,0 @@
-"""AI summarization module.
-
-Supports OpenAI-compatible (doubao, deepseek, openai) and Anthropic APIs.
-Uses urllib.request for zero extra dependencies.
-"""
-
-from __future__ import annotations
-
-import json
-import logging
-import ssl
-import urllib.request
-from typing import Any, Dict, List
-
-from .models import Tweet
-
-logger = logging.getLogger(__name__)
-
-SYSTEM_MESSAGE = "你是一个专业的 Twitter/X 信息流分析师，擅长提炼关键信息和发现趋势。"
-
-
-def _build_prompt(tweets, language="zh-CN"):
-    # type: (List[Tweet], str) -> str
-    """Build the summarization prompt."""
-    lines = []
-    for i, t in enumerate(tweets):
-        score_str = " [score: %.1f]" % t.score if t.score else ""
-        rt = " (RT by @%s)" % t.retweeted_by if t.is_retweet and t.retweeted_by else ""
-        media_str = ""
-        if t.media:
-            media_str = " [%s]" % ", ".join(m.type for m in t.media)
-        url_str = ""
-        if t.urls:
-            url_str = "\n   Links: %s" % ", ".join(t.urls)
-        quoted = ""
-        if t.quoted_tweet:
-            qt = t.quoted_tweet
-            quoted = "\n   Quoting @%s: %s..." % (qt.author.screen_name, qt.text[:100].replace("\n", " "))
-
-        text_preview = t.text.replace("\n", " ")[:300]
-        lines.append(
-            '%d. @%s (%s)%s%s\n'
-            '   "%s"\n'
-            '   ❤️%d 🔄%d 💬%d 🔖%d 👁️%d%s%s%s'
-            % (
-                i + 1, t.author.screen_name, t.author.name, rt, score_str,
-                text_preview,
-                t.metrics.likes, t.metrics.retweets, t.metrics.replies,
-                t.metrics.bookmarks, t.metrics.views,
-                media_str, url_str, quoted,
-            )
-        )
-
-    tweet_summaries = "\n\n".join(lines)
-
-    if language.startswith("zh"):
-        lang_inst = "请用中文输出。"
-    else:
-        lang_inst = "Please output in %s." % language
-
-    return (
-        "你是一个 Twitter/X 信息流分析师。请对以下 %d 条推文进行摘要总结。\n\n"
-        "要求：\n"
-        "1. 按主题分组（如：AI & 编程、Crypto、工具推荐、生活观点等）\n"
-        "2. 每组列出关键推文和核心观点，标注作者 @handle\n"
-        "3. 标注数据亮点（高赞/高收藏推文用 🔥 标记）\n"
-        "4. 最后用 2-3 句话总结今天 timeline 的整体趋势\n"
-        "5. %s\n\n"
-        "推文数据：\n\n%s"
-    ) % (len(tweets), lang_inst, tweet_summaries)
-
-
-def _call_openai(prompt, config):
-    # type: (str, Dict[str, Any]) -> str
-    """Call OpenAI-compatible API."""
-    url = config.get("base_url", "").rstrip("/")
-    if not url.endswith("/chat/completions"):
-        if not url.endswith("/v1"):
-            url += "/v1"
-        url += "/chat/completions"
-
-    payload = json.dumps({
-        "model": config.get("model", ""),
-        "messages": [
-            {"role": "system", "content": SYSTEM_MESSAGE},
-            {"role": "user", "content": prompt},
-        ],
-        "temperature": 0.3,
-        "max_tokens": 4096,
-    }).encode("utf-8")
-
-    req = urllib.request.Request(url, data=payload)
-    req.add_header("Content-Type", "application/json")
-    req.add_header("Authorization", "Bearer %s" % config.get("api_key", ""))
-
-    ctx = ssl.create_default_context()
-    with urllib.request.urlopen(req, context=ctx, timeout=120) as resp:
-        data = json.loads(resp.read().decode("utf-8"))
-
-    choices = data.get("choices", [])
-    if choices:
-        return choices[0].get("message", {}).get("content", "")
-    return ""
-
-
-def _call_anthropic(prompt, config):
-    # type: (str, Dict[str, Any]) -> str
-    """Call Anthropic Messages API."""
-    url = config.get("base_url", "").rstrip("/")
-    if not url.endswith("/messages"):
-        if not url.endswith("/v1"):
-            url += "/v1"
-        url += "/messages"
-
-    payload = json.dumps({
-        "model": config.get("model", ""),
-        "system": SYSTEM_MESSAGE,
-        "messages": [{"role": "user", "content": prompt}],
-        "temperature": 0.3,
-        "max_tokens": 4096,
-    }).encode("utf-8")
-
-    req = urllib.request.Request(url, data=payload)
-    req.add_header("Content-Type", "application/json")
-    req.add_header("x-api-key", config.get("api_key", ""))
-    req.add_header("anthropic-version", "2023-06-01")
-
-    ctx = ssl.create_default_context()
-    with urllib.request.urlopen(req, context=ctx, timeout=120) as resp:
-        data = json.loads(resp.read().decode("utf-8"))
-
-    content_blocks = data.get("content", [])
-    for block in content_blocks:
-        if block.get("type") == "text":
-            return block.get("text", "")
-    return ""
-
-
-def summarize(tweets, config):
-    # type: (List[Tweet], Dict[str, Any]) -> str
-    """Summarize tweets using the configured AI provider.
-
-    Config keys: provider, api_key, model, base_url, language
-    """
-    api_key = config.get("api_key", "")
-    if not api_key:
-        raise RuntimeError(
-            "AI API key not configured.\n"
-            "Set ai.api_key in config.yaml or export AI_API_KEY=your_key"
-        )
-
-    if not tweets:
-        return "No tweets to summarize."
-
-    language = config.get("language", "zh-CN")
-    prompt = _build_prompt(tweets, language)
-    provider = config.get("provider", "openai")
-
-    logger.info("Calling AI (%s/%s)...", provider, config.get("model", ""))
-
-    if provider == "anthropic":
-        return _call_anthropic(prompt, config)
-    else:
-        return _call_openai(prompt, config)