> For clean Markdown of any page, append .md to the page URL.
> For a complete documentation index, see https://docs.beatapi.io/llms.txt.
> For AI client integration (Claude Code, Cursor, etc.), connect to the MCP server at https://docs.beatapi.io/_mcp/server.

# How do I retrieve recent Bilibili comments to review new feedback?

> Use time ordering for new feedback and popular ordering for highly engaged discussion, keeping targets, observation times and original wording separate.

For recent Bilibili comments, call `bilibili.app.fetch_video_comments` with `mode:2`. For popular comments, run a separate `bilibili.app.fetch_video_comments` lookup with `mode:3`. After a course update, read the recent comments to find questions about the new chapter, and read the popular comments to see which issues already drew discussion.

## Retrieve the first recent-comment page

`data:bilibili.app.fetch_video_comments` accepts a BV ID or a confirmed AV ID. Replace the BV ID in the request below; the first request uses `next_offset:1`:

```bash
curl -sS https://api.beatapi.io/v1/social-data/call \
  -H "Authorization: Bearer $BEATAPI_API_KEY" \
  -H 'Content-Type: application/json' \
  -d '{"action":"bilibili.app.fetch_video_comments","params":{"bv_id":"<bv_id>","mode":2,"next_offset":1}}'
```

With only a video link, identify the video with the [BV/AV helper](/faq/bilibili-bv-comments). The Web action `data:bilibili.web.fetch_video_comments` takes `bv_id` and `pn` and has no App `mode` input.

## Choose the ordering for the task

| Goal                                         | Ordering                  | Review                                                  |
| -------------------------------------------- | ------------------------- | ------------------------------------------------------- |
| Find questions after a video update          | Time, `mode:2`            | New questions, applicable version and existing answers  |
| Investigate discussions with more engagement | Popular, `mode:3`         | Original comment, replies and reasons to investigate    |
| Compare the observations                     | Save both runs separately | Newly observed questions versus established discussions |

Use popular comments to discover discussion leads. A high like count does not make a comment correct or representative of every viewer. Keep the original wording, your summary and the next actions in separate columns.

## Organize the returned feedback

1. **Save the run.** Store the video target, `mode`, collection time, raw response and `request_id`.
2. **Verify the comments.** Check the actual comment text, identifiers and times in the raw response.
3. **Build the feedback rows.** Record question, excerpt, source and action needed for each comment you keep.
4. **Merge repeats.** When a comment appears again, reuse the existing comment record and add the new observation time.

For an assistant connected to BeatAPI:

> Retrieve one recent-comment page for this course video and look for questions about the new chapter. Use actual returned wording, with video, query time and sources. Mark points already explained in the video and list the rest for follow-up.

The comment output has no verified fixed list path or continuation path yet, so this page covers the first-page request and a manual review method. Automatic pagination, retrieving exactly 50 comments and an automatic topic table require field mapping first. Record the actual number of comments each page returns.

## Review again later

1. Start each new time-ordered observation from page 1 and save the new observation separately from earlier runs.
2. When switching between popular and recent ordering, start again from page 1; do not continue with the cursor from the other ordering.
3. For reply context, verify the reply route for the selected parent comment separately.

See current [pricing](https://beatapi.io/pricing). To revise your analysis, reclassify the saved feedback instead of retrieving the comments again.

[Bilibili questions](/faq/bilibili)