Social Media Copilot: a hobby-built scraping extension that gathered 40k users over two years, was delisted, then climbed back to 10k from 0

Social Media Copilot is a Chrome extension that batch-scrapes posts, comments and creator data from Xiaohongshu, Douyin, Kuaishou, Bilibili and TikTok pages, then exports to Excel or syncs straight into Feishu Bitable. In the store it looks like a new product that hit 10k users in three months, but it's actually an old open-source extension that started 2024 years ago. The old listing reached 40k users, then vanished from the store, and the author published a new listing and brought the old users back. Below: how it grew, where the money comes from, who's competing, and whether you should follow it.

A Monday-night post on V2EX

On Dec 2024 2 evening, in V2EX's "Share Your Creations" thread, a user named AndyZh posted: "Wrote a Chrome extension in my spare time that collects data and materials from social platforms like Douyin and Xiaohongshu. Open source." The post was brief: it only supports Xiaohongshu and Douyin, exports watermark-free images and videos, batch-exports comments, and batch-exports all notes from a given author. It works by injecting scripts into the page and calling the platforms' own APIs. The closing line warns, "The extension has no rate limits, so heavy scraping can easily get accounts banned. Please control your frequency," adding "Solo development means limited bandwidth, so updates are slow."

The post got 33 replies (checked on V2EX, 11-06). Some said "Exactly what I needed, thanks," others reported "Xhs comment export keeps failing" and "Douyin video download errors," and the author replied to each. The most interesting one: someone posted a screenshot saying they couldn't install it. The author replied: "I did something dumb. I tried to cancel the review, and it canceled the publish instead. For now Chrome only allows installing the crx from GitHub, dragged in manually. 😭"

At that point it had 976 users and 8 ratings in the Chrome store (from web archives, 2024-12-03).

Then came a fairly flat growth curve

It didn't break out. It built up gradually:

Source: web archives of the old listing. The new listing is from the Chrome store; from 11-01 to 11-06 it stayed at 10k users and 41 ratings, with no change.

2026 Jul 22: the update log mentions "launching the new offline installer"; Jul 27: developer xisence published the current listing; by late August, the old listing had only a generic "Chrome Web Store" title left. Why the delisting wasn't caught is unclear: either Google took action or the author moved on purpose, and the outside view can't tell which. But the timeline fits "old listing got into trouble, so the new one was set up" (inference). The 10k from this new listing over the last three months came mainly from users redirected by the website and open-source README, not from new acquisition.

Also, the store numbers understate the reach. The official site itself says "The Chrome Web Store is usually not directly accessible from mainland China." Download pages and help docs push people to install the offline package. Those users don't show up in store stats, and nobody knows how many there are.

What it is, and why people stay

In one line: China-based operators click a few times in their browser and get a batch of notes, videos, comments and creator info as a spreadsheet.

It survives among many downloaders thanks to three things:

First, it keeps getting fixed. From v0.0.1 (2024-09) to v3.6.2 (2026-11-04), the changelog has over sixty versions, most of them "fixed a Xiaohongshu / Douyin feature." The latest versions read "optimized Douyin scraping retry and risk-control prompts" and "improved RPA scraping stability." That's the whole game for this kind of tool: when a platform redesigns, it breaks, and users stick with whoever keeps repairing it.

Second, more platforms, and it plugs into ad-buying back-ends. It started with only Xiaohongshu and Douyin, then added Kuaishou, Bilibili and TikTok, plus creator scraping and bulk outreach for Xingtu (Douyin creator ads) and Pugongying (Xiaohongshu creator ads). The latter is a workflow brands and MCNs actually pay for.

Third, data goes to Feishu sheets. In August 2025, v2.0 added Feishu Bitable sync, with append, merge-update and overwrite modes. Combined with scheduled scraping, the extension went from a "downloader" to a pipe that maintains content and creator databases. After v2.0, the old listing grew from about 10k to 40k (correlated in time, causation not verified). Overseas similar extensions don't do Feishu, and few Chinese small extensions do it well.

The website touts "runs locally" as a selling point: it scrapes "pages you can see" using your own account in your own browser, so you don't hand over account credentials. When risk controls kick in, you can fall back from "API mode" to "RPA mode," which simulates a person clicking and refreshing.

Where the money comes from

Basic features are free, no sign-up. VIP is ¥12–15/month per platform, or ¥29/month for all platforms (socialext.com/pricing, 11-06). The difference lies in multi-account, RPA mode, unlimited Feishu sync and resumable scraping: limits that "only heavy users hit."

Before there was a payment system, the author tested with "sponsorship × 5 = VIP days." On the sponsor page: 69 individuals, ¥2,169 in total, ¥31 per person on average, mostly ¥8.88 and ¥18.88. The comments show who the users are: "So good, and free, great job," "Brother, you have to keep going!" "Thanks! Please keep updating, I'd pay for a paid version too," "Great tool, I run a self-media account called AI Geng, I'll help promote it," and one person who sponsored ¥188.88 wrote, "Good, support it. Feels like I missed an opportunity, brother."

How many people buy VIP now isn't public. At ¥29/month and a base of tens of thousands of users, this is a small business that could support one or two people, not a big one (inference).

There's also a counter-signal that the demand is real: the official site has a dedicated "third-party paid download notice," saying people are charging for installers of this free extension through Baidu search ads and trading platforms.

Where users come from

• GitHub open source plus V2EX, confirmed. The open-source repo social-media-copilot now has about 1,400 stars and 229 forks. The first batch of nearly a thousand users came from that V2EX post. The open-source version stayed up, and its description, "full-featured version," points to the new store listing, funneling people toward the paid version.
• Word of mouth, small scale. The WeChat Official Account "Light Up Your Full-Stack Skill Tree" wrote a tutorial, and Smzdm (what's worth buying) reposted it 2026-04-26 times. On Xiaohongshu there are notes like "Social Media Copilot: a Douyin and XHS data-scraping godsend!" (36 likes) and "0 basics! Social Media Copilot install and full scraping walkthrough," with likes in the dozens. A bunch of AI tool directory sites list it. Nothing on X, no media coverage.
• Store search (inference). The title is literally "Douyin, Xiaohongshu data scraping tool," so it catches searches for those terms.
• No trace of it running ads was found.

Who it's competing with

This space is crowded, but most players only took one small slice of it.

Scale comes from the Chrome store and GitHub, around 11-06.

Verdict: the one that does multi-platform + structured data + Feishu + creator-deal backend all at once is the biggest. But that moat was built by fixing things every week for two years. It's not a single idea.

Its biggest risk

It sits in the gap between platform anti-scraping and Chrome Web Store review. The old listing with 40K users disappeared overnight. The author's response: offline installers, and split-out "open-source edition" (about 570 users) and "Douyin & Xingtu edition" (about 132 users) listings to spread the risk (third-party plugin sites data). That setup itself shows the author knows the next takedown could come at any time.

If it were you

One-line take: The demand is real and people will pay, but this is a small business that only survives if you keep fixing anti-scraping for two years. The top spot is already held, so following it isn't worth much. The better thing to learn is its playbook.

Who it suits: One or two people already building domestic ops / MCN tools, who know each platform's interfaces and are willing to maintain it long-term; or someone who only targets the one platform that social helpers handle poorly.

Who it doesn't suit: Anyone wanting a quick win from a single idea; anyone who can't handle the store delisting them or platforms changing without warning; anyone hoping to make big money from it. At ¥29/mo the ceiling is low.

How to approach it: Don't build a second all-platform scraper. What's worth trying is where it doesn't cover or does shallow work: Weibo, WeChat Channels, WeChat Official Accounts (sponsor comments already ask for Weibo). Or move one step downstream: scraping is just the entry point, and what you sell is the creator database / competitor monitoring sheet itself, e.g. a daily auto-updated creator table for one category. An overseas version is also possible: the same "scrape TikTok / Instagram / YouTube into Google Sheets / Airtable / Notion", but KOLSprite already occupies that, so research the competitors first.

How to get users: copy it. Open-source a working v0 first, post on V2EX, Juejin, and GitHub. Ship a free, no-signup base version and let the ops folks write tutorials on Xiaohongshu (RedNote). Once users pile up, put the limits heavy users hit (multiple accounts, unlimited sync, resumable scraping) behind VIP.

Time pressure: This isn't a window-type opportunity with a deadline. But the later you start, the harder it gets to catch up with the leader's version iterations and reputation.

Signs you've misjudged it: If the new listing climbs back toward 30K and 40K over the next few months, while followers like MeiXun Assistant also grow, the market is bigger than it looks and can fit a second player, so reassess. Conversely, if the new listing gets delisted again, or Xiaohongshu and Douyin fully tighten web-side interfaces, this path is basically closed.