Add WeChat and Xiaohongshu research feeds to TrendRadar
Add two subscriptions to rss.feeds in TrendRadar’s config/config.yaml: one feed for new WeChat articles and one feed for Xiaohongshu leads. Fetch account listings or search results with BeatAPI, generate an RSS file from the selected items (for Xiaohongshu with make_feed.py below), and publish each RSS file at an HTTPS URL that TrendRadar can reach. Add the first feed and get one sample item into the TrendRadar digest before adding the second feed.
Prepare two reachable feeds
- WeChat: Build a candidate list with incremental article checks. For selected articles with existing text records, reuse the local RSS generator.
- Xiaohongshu: Get note IDs, titles and search excerpts with the search export. Select the notes, and add an independently verified post link to each selected record. Fetch note details only when the saved material falls short.
A title-and-link feed reuses the saved material; no article bodies are needed. In your RSS publishing step:
- Create one XML file per source.
- Give each item a real
titleandlinkplus a stableguid: the account and article key for WeChat, the note ID for Xiaohongshu, with a distinct prefix per source. A repeated observation of the same post keeps the sameguid. - Put any saved excerpt in
description. - Write
pubDatein RSS date format only when you know the real publication time. - Verify one real item, then serve the XML files at HTTPS URLs reachable from the TrendRadar runtime.
Generate Xiaohongshu RSS from the reviewed shortlist
Save the reviewed items in notes.json. Set link to the feed homepage and each items[].url to a verified original post. Provide a timezone-aware ISO 8601 published_at when the publication time is known, and omit published_at otherwise. Replace the example title, URLs and ID.
Save the script as make_feed.py and run python3 make_feed.py notes.json xiaohongshu.xml with Python 3.8 or later. make_feed.py replaces xiaohongshu.xml only after the new XML parses. Empty input, duplicate IDs and invalid dates leave the previous xiaohongshu.xml intact; correct notes.json and rerun make_feed.py locally.
Add the feeds in TrendRadar
Merge the two entries below into rss.feeds in your deployment’s config/config.yaml. Keep the other feeds already in rss.feeds, and replace the example URLs. Inside a container, localhost refers to the TrendRadar container itself, not to the host.
- From the TrendRadar runtime, verify that each feed URL can be fetched.
- Run one TrendRadar cycle and inspect the retrieval log and the report.
- In keyword mode, add a term the test item should match to
config/frequency_words.txt. - Confirm that
display.regions.rssis enabled and thatdisplay.region_orderincludesrss, then check the age filter. - Schedule the feed update so that feed generation finishes before TrendRadar fetches.
Diagnose a missing report item
For example, for a digest about “cold brew,” put two same-day items in a test feed: “Cold brew coffee for beginners” and “How to clean an oven.” With keyword filtering enabled and “cold brew” configured, the retrieval log should show the test feed fetched successfully, and the digest should show “Cold brew coffee for beginners” with the original link. If both items are in the test feed but the coffee item is absent from the digest, check the keyword and display settings. Then publish the coffee item again with the same guid, and confirm that the notification mode treats the coffee item as an existing item.
When a run has no approved notes, make_feed.py keeps the previous XML. Log the feed name, run time and result, for example “Xiaohongshu shortlist, Tuesday 09:00, no approved notes, Monday’s file kept.” If the log shows a new file but TrendRadar still sees Monday’s items, check from the TrendRadar runtime which file the feed URL actually serves. When editing a title, keep the same note ID and guid.
Checking one page for each of five WeChat accounts and three Xiaohongshu keywords takes eight data requests per day. Regenerating the XML from saved records requires no new lookup. Extra pages and note details are billed per request. BeatAPI pricing
TrendRadar RSS config example · TrendRadar feeds and keyword filters · RSS 2.0 specification

