Search a WeChat Official Account by name. Returns candidates with fakeid.

Use the returned fakeid as input for list_articles / crawl_incremental.

Fetch one page of articles for a given fakeid and store metadata.

Use this for a first-time full crawl: call repeatedly with begin += count until the returned articles list is empty. The metadata is upserted into the local SQLite store; bodies are NOT fetched here.

Pull only newer articles than what is already stored locally.

Walks pages from newest to older, stopping once a page contains no articles newer than the latest stored update_time, or when max_pages is reached.

Anti-detection: page size varies per call (3-6), delays are jittered, and occasional 30-90s pauses are inserted. Calls are gated to local work hours (env: WECHAT_MP_MCP_WORK_HOURS, default 8-23). Pass override_work_hours=True to bypass the gate for urgent runs.

Fetch an article's full body, parse to Markdown, and optionally cache it.

url must be a https://mp.weixin.qq.com/s/... link. If save=True and the article exists in the local store, the markdown body is written back.

Report today's API call usage against the daily safety cap.

Only counts calls that hit the mp.weixin.qq.com backend (searchbiz + appmsg). Reading public article URLs does not count.

List articles already stored locally for a given fakeid, newest first.