> 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.

# Is WeChat history complete, and can stopping at an old article miss new ones?

> Separate exhausted visible history, errors and scan limits, without assuming strict chronological ordering or complete historical coverage.

A WeChat history export is complete only up to the point where the scan stopped, so report coverage with **the account, tab, page count and stop reason**. If a successful response from `wechat_mp.v2.fetch_account_articles` reports `data.is_end=1`, say you reached the end of the list visible to this lookup. If your page limit came first, report the range you actually checked. The list does not guarantee strict chronological order, so an older article on a page is not a stop signal.

## Put the coverage in one line of the report

For example, the target was six months, but you checked three pages and the earliest article found is dated October 12. Write: “Three article pages checked. 26 unique articles observed from October 12 to November 17. Stopped at the page limit.” Keep that scope when you select examples. To compare publishing frequency over six months, fetch more pages until the intended period is covered.

| Why the scan stopped                   | What to report                                                                    |
| -------------------------------------- | --------------------------------------------------------------------------------- |
| The response marks the end of the list | Reached the end of the visible list. Keep the final response and collection time. |
| Page or cost limit reached             | Checked up to the limit. The remaining range is pending.                          |
| Cursor is missing or repeated          | Paging stopped. Keep the last successful response and cursor.                     |
| A request failed                       | Scan incomplete. Keep the range already checked and the error record.             |

## Where should a daily update start?

Start each daily scan on page one and read `data.articles` with `raw:false`. Compare articles by account, `app_msg_id` and `idx`, so that each article in a publication batch stays separate. The first page can include articles you already saved; keep paging with the current scan's cursor within the chosen page range.

[Daily update workflow](/guides/wechat-incremental-articles) · [Historical export](/guides/wechat-account-articles)

## What does a replay or failed page do to saved history?

The daily-update processor writes a new state file and never overwrites the previous state file:

* Replaying saved pages against the state they produced adds zero new IDs.
* An existing ID with a changed title or other list metadata goes into `changed-articles.csv`, not `new-articles.csv`.
* A failed tail page keeps the validated earlier pages and records `gateway_failed`.
* An empty page does not delete old articles.

## How does the history checkpoint become a daily scan?

The [history helper](/guides/wechat-account-articles) writes two files: a checkpoint for continuing the same backfill, and an account-scoped `state.json`. Keep every history chunk as source evidence. The daily helper reads `state.json` but starts with an empty `offset`; the daily helper does not use the history checkpoint. If an old cursor is rejected, the explicit `--restart` flow begins at page one and keeps the known article identities. An initial backfill that stopped at its budget stays partial, even after a successful daily scan.