ShiPinHao Scraper
Social MediaConnect AI agents to WeChat Channels (ShiPinHao), China's short video platform. Agents look up video accounts, read individual videos, and search public content for monitoring and research.
What This Integration Enables
WeChat Channels, known in Chinese as ShiPinHao, is one of the largest short video platforms in the world and one of the least accessible to Western tooling. It publishes no general purpose API for reading public content, which is why almost every operations stack has a blind spot exactly where a significant consumer market is. This connector fills that gap through TikHub, a credit metered data API, and it is worth being straightforward about what that means: the API key you configure is a TikHub key, not a WeChat one, and the credits it consumes are the real cost model of running this. Get API Account Info returns the account tied to that key with its remaining credit balance and status, which is the first call any scheduled flow should make.
Every action here is a read. There is no publishing path, no reply path, no moderation path, and nothing this connector does changes anything on WeChat Channels. That constraint is what makes it usable as a monitoring surface rather than a marketing surface. Search Videos takes a keyword and returns matching videos and accounts, which is where a flow starts when it does not already have an identifier. Get Account Profile takes either a finder username or a finder ID and returns the account's nickname, avatar, signature, and follower statistics. List Account Videos returns the works an account has published, paginated by passing the cursor from the previous response. Get Video Detail takes a post ID, also called the object ID or export ID, and returns the full record: caption, media, author, and engagement metadata.
There are no triggers, and there is no polite version of that. WeChat Channels does not notify anyone, TikHub does not push, and nothing about this connector wakes a flow up when a video is posted. Monitoring here is polling by construction: a schedule runs the searches, the results are diffed against what was seen last time by post ID, and the new ones are what the flow acts on. That is a perfectly good pattern and it is worth designing for deliberately, because the diff state has to live somewhere the flow can read on the next pass, and because the credit balance means the polling interval is a budget decision rather than a free one.
Without FlowRunner
With FlowRunner
Use Case Scenarios
Watching a market nobody else on the team can see
A brand selling into China wants to know what is being said about it on WeChat Channels without asking someone to search by hand each week. A daily flow calls Get API Account Info to confirm there are credits, then runs Search Videos across a fixed set of keywords: the brand name in Chinese and in Latin script, the flagship product names, and the common misspellings. For each result it has not seen before, it calls Get Video Detail for the caption and engagement figures and Get Account Profile for the publishing account's follower count. Everything new lands in Google Sheets with the date first seen, and anything above an engagement threshold is posted to Slack with the caption translated alongside the original. The team's picture of the market stops depending on who happened to look.
Following the creators you already work with
A campaign runs through a set of named creators, and the reporting obligation is to know what each of them actually published and how it performed. Rather than searching, the flow works from identifiers. For each creator it calls Get Account Profile with the finder username to confirm the account and capture the current follower count, then List Account Videos paginated with the cursor from the previous response until it reaches videos it has already recorded. Each new video goes through Get Video Detail for its engagement metadata. The result is written to a campaign record in Airtable with the follower count at the time of publication beside each video, so a performance comparison months later is against the audience the creator actually had then rather than the one they have now.
A competitive read that runs whether or not anyone remembers
Understanding a competitor's content cadence on a platform you cannot log into is otherwise a manual research task that gets done once and then never repeated. A weekly flow calls List Account Videos for each competitor account, records posting frequency, format, and the engagement distribution across their recent works from Get Video Detail, and writes the series to a sheet. The interesting output is not any single video. It is the shape over time: whether a competitor increased cadence before a launch, which formats they abandoned, and where their engagement is concentrated. Because the flow also calls Get API Account Info each run and reports the remaining balance, the research keeps running instead of stopping silently when the credits run out mid-quarter.
Human-in-Loop Highlight
This connector cannot act on WeChat Channels, and that is precisely why the decision it feeds needs a person rather than a threshold. The realistic high stakes use is brand protection: Search Videos surfaces content that looks like counterfeit product, unauthorized use of brand assets, or a distributor selling outside its territory, and the next step is a takedown request or a legal letter filed by a human somewhere else entirely. Nothing in this surface files it, reverses it, or softens it. What the agent has is a keyword match and some public metadata, and a keyword match is not proof. A creator legitimately reviewing a product, a reseller with a valid agreement nobody told marketing about, and a genuine counterfeiter can all return from the same search with similar captions, and the difference between them is exactly the kind of context that lives in a contract folder rather than in an API response. Getting it wrong means accusing a named individual with a real audience of selling fakes. So the agent assembles rather than concludes. It calls Get Video Detail for the caption, media, and engagement figures, Get Account Profile for the account's nickname, signature, and follower statistics, and List Account Videos to show whether this is an established creator or an account that appeared last week, then puts the package to the brand protection owner: "Three new matches for the product name this week. Account A, 240,000 followers, posting product content since 2024, caption reads as a review. Account B, 900 followers, created 11 days ago, six near-identical videos with a purchase link. Account C is an existing distributor. Escalate any of these for takedown?" The owner reads what the flow cannot, which is the relationship. Treating this as a human-in-the-loop escalation rather than an automatic rule is what makes daily monitoring safe to leave running, and the credit balance from Get API Account Info is reported alongside it because the searching itself has a budget.
Agent Capabilities
5 actionsAccount
1- Get API Account Info Retrieves the current TikHub account associated with the configured API key, including the remaining credit balance and account status. Use it to verify the key is valid and to check how many request credits are left before running a scraping pass, since every other action here consumes them.
Videos
1- Get Video Detail Retrieves the full detail of a single WeChat Channels video by its post ID, including caption, media, author, and engagement metadata. The post ID is also called the object ID or export ID. This is the action that turns a search hit into evidence.
Accounts
2- Get Account Profile Retrieves the profile of a WeChat Channels video account, including nickname, avatar, signature, and follower statistics. Identify the account by either its finder username or its finder ID; at least one is required.
- List Account Videos Retrieves the videos an account has published. Identify the account by finder username or finder ID, at least one required. Results are paginated: pass the cursor returned by the previous call to fetch the next page, and stop when the results reach videos already recorded.
Search
1- Search Videos Searches WeChat Channels content by keyword and returns matching videos and accounts. This is where a flow starts when it does not already hold a post ID or a finder ID, and it is the action that makes standing keyword monitoring possible on a platform that publishes no feed.
Frequently Asked Questions
What can FlowRunner do with ShiPinHao Scraper?
FlowRunner agents can run Get API Account Info, Get Video Detail, and Get Account Profile in ShiPinHao Scraper, plus 2 more actions.
Does connecting ShiPinHao Scraper to FlowRunner require OAuth?
No. ShiPinHao Scraper connects to FlowRunner with an API key, no OAuth flow required.
Can ShiPinHao Scraper trigger a FlowRunner workflow automatically?
ShiPinHao Scraper doesn't currently expose triggers in FlowRunner. It connects as an action step inside workflows started by another trigger.
Start building with ShiPinHao Scraper
$100 in credits. No card required. Connect in minutes.