twscrape says No active accounts and my X accounts keep getting locked. What now?

The usual story: a nightly job pulls posts for 40 keywords through twscrape, X locks two accounts, the rest hit their limits, and the log ends with No active accounts. Buying more accounts and proxies buys a few weeks. If the job only reads public posts and profiles, you can drop the pool and call an API with a key instead. Nothing logs in to X on your side, so there is nothing to lock.

What maps to what

twscrape	okdata endpoint	Price
api.search(query)	X post search	$0.006 per page
api.user_tweets(uid)	X account posts	$0.006 per page
api.user_by_login(login)	X account profile	$0.001
api.tweet_details(id)	X post detail	$0.001

Search takes the same query you would type into X, operators included, and sort is latest or top. Account calls take the handle, not the numeric user ID.

The search loop, rewritten

import os, requests

API = "https://api.okdata.online/v1"
H = {"Authorization": f"Bearer {os.environ['OKDATA_API_KEY']}"}

def search(query, pages=5):
    cursor, posts = None, []
    for _ in range(pages):
        params = {"query": query, "sort": "latest"}
        if cursor:
            params["cursor"] = cursor
        body = requests.get(f"{API}/x/posts/search", headers=H, params=params, timeout=60).json()
        if body["code"] != "ok":
            raise RuntimeError(f"{body['code']}: {body['message']}")
        posts += body["data"]["items"]
        if not body["page"]["has_more"]:
            break
        cursor = body["page"]["next_cursor"]
    return posts

for p in search('"vector database" -filter:replies lang:en'):
    print(p["published_at"], p["author"]["username"], p["metrics"]["likes"], p["text"][:80])

Each post has text, published_at, url, lang, the author with follower count, and metrics with views, likes, replies (comments), reposts and bookmarks (favorites).

What a nightly job costs

40 keywords at 3 pages each is 120 calls a night: {{cost:x.posts.search120}}, or about {{cost:x.posts.search3600}} for a 30-day month. Each page is a separate charge, failed calls are free, and a rate_limited reply tells you how many seconds to wait. The retry rules are on the errors page.

What you give up

• Follower and following lists, likers, retweeters and list timelines are not in the catalog. twscrape gets those, if you keep accounts alive.
• There is no reply-thread endpoint. Replies to an account can be found with a to:handle search, one page at a time.
• You see what X search shows to the public: no protected accounts, nothing deleted.

If you need the graph data or a very large daily volume, a carefully paced twscrape pool or X's own API is still the better fit. For a keyword or account feed, a key is less to babysit. The word-of-mouth guide shows a full monitoring run.