Skip to navigation

How do I find questions to answer or verify in replies to an X announcement?

To find open questions under an X announcement, fetch the announcement’s replies with twitter.web.fetch_post_comments and the announcement’s tweet_id. Compare each reply with the announcement text and mark each question as answered, needing clarification or needing a product check. Hand over the exact reply, the relevant region or version, the source and the next step.

Use the announcement as the reference

Save the announcement’s link, post ID and the text or version you actually read. A reply asking “When does it start?” might concern general availability, one region or a waiting list. If you group these replies as one question without checking the announcement, the follow-up answer can miss the point.

If you already have the announcement text, reuse the text. Retrieve the original post with a separate detail action only when you need the text; the detail action has its own contract and cost. A short preview does not show that you read every clause of the announcement.

What should I decide for each reply?

Use the table as a reading guide for common reply situations:

Reply situationChecklist decisionNext step
A question about an unstated eligibility ruleOutstanding question, with exact quote and relevant clauseAsk the product owner to verify before drafting clarification.
Something the announcement already answersAnswered, but perhaps hard to find; record the source clauseConsider making the explanation easier to locate, without counting it as a new feature need.
A claim of failure without stepsNeeds verification, not a confirmed defectIdentify missing version, conditions or reproduction steps for support review.
A response to someone else’s experience or sarcasmContext unresolvedLocate the parent discussion; do not infer a parent relationship from usernames.
Repeated versions of one questionPreserve each source before grouping a themeOccurrences are not unique people or the share of all readers.

Likes help you decide which replies to read first. Likes do not show that a reply is correct, that the announcement is wrong or that most readers agree. The displayed total reply count does not reveal what uncollected replies say.

Hand over an answer checklist

For each reply, record by hand: the announcement ID, the reply source, the exact quote, the summarized question, the relevant announcement clause, the status (answered, outstanding or unverified), the missing context, a suggested owner and the next action. Keep quotes and summaries in separate columns. When a reply ID, parent relationship or text is unavailable, leave the cell blank and write the reason; do not invent the value.

From the checklist, a teammate can see which questions need clarification, which need a factual investigation and which already have an answer. A suggested owner is not an assignment, and a drafted answer is not posted automatically. The feedback handoff principles help you organize the follow-up work, but the brand-mentions script accepts only the brand-mentions guide’s search export; the script cannot read a reply JSON of unknown shape.

Bound the scope and know when to stop

Start with one page, and investigate the useful questions before collecting more. Continue only with an actual cursor that you confirmed in the response, for the same announcement. Each further page, detail lookup and context lookup is another call. Stop when the response has no continuation, when a page repeats, when an error occurs or when the approved budget is used; do not retry automatically to get complete coverage.

In the handover, state the observation time, the pages actually read and the parts not reviewed. The checklist covers only the replies actually read; the checklist is not complete discussion coverage and not continuous monitoring. Steps 1–5 below give the input and retrieval steps.

1. Identify the parent post

In https://x.com/<handle>/status/<tweet_id>, the value after /status/ is the post ID. Keep the full URL as the source, and store the post ID as a string so that spreadsheets and numeric types do not lose precision. An author handle is not a tweet ID.

If you only need the replies and already know the post ID, skip the detail call. Fetch the parent post’s details separately only when the analysis needs the parent post’s text, and count the detail call as another execution.

2. Inspect the post-replies action

curl -sS https://api.beatapi.io/v1/capabilities/inspect \
-H 'Content-Type: application/json' \
-d '{"reference":"data:twitter.web.fetch_post_comments"}'

Inspect is free. twitter.web.fetch_post_comments requires tweet_id and accepts a string cursor. Do not substitute fetch_user_tweet_replies: fetch_user_tweet_replies takes screen_name and returns an account’s replies, not the replies to your target post.

3. Retrieve the first page

Check the current price, then send the first-page request:

curl -sS https://api.beatapi.io/v1/social-data/call \
-H "Authorization: Bearer $BEATAPI_API_KEY" \
-H 'Content-Type: application/json' \
-d '{"action":"twitter.web.fetch_post_comments","params":{"tweet_id":"<tweet_id>"}}' \
-o x-post-replies-1.json

Check the HTTP status and the execution status, then keep the raw file x-post-replies-1.json. twitter.web.fetch_post_comments is currently runnable without a published fixed output schema. Search-action paths such as data.timeline and data.next_cursor are not a contract for replies. Locate the reply list and the continuation cursor in the actual response first.

4. Bound pagination

Pass the actual returned cursor into the next request, and keep the same tweet_id. twitter.web.fetch_post_comments publishes no sorting input: do not add sort, and do not promise newest-first or most-liked ordering. If you need another ordering, inspect an alternative action separately; the alternative action’s results are not the next page of twitter.web.fetch_post_comments.

Set a cap such as two pages. Stop when the continuation is missing or repeats, when an error occurs or when the approved budget is reached. Each page is billed separately. The displayed reply total can differ from the number of records currently retrievable; the difference does not reveal the content of unavailable replies.

5. Preserve context in the handoff

ColumnContent
ParentOriginal URL and ID; quote parent text only if actually read.
Reply sourceReal reply ID, author and returned link when available.
QuestionSeparate the short quote from your interpretation.
ContextRelevant announcement clause and any missing parent reply.
Next stepClarify, investigate or review manually; do not auto-post.

Record an allegation in a reply as a claim to verify, not as an established product defect. When thread context is missing, say so in the Context column; do not replace the missing context with an inferred opinion.

Inspect the X post-replies action and price. After approval, fetch one page for my announcement post. Preserve available sources, distinguish quotes from interpretation and list questions needing answers. Do not use the account-replies timeline, paginate or reply publicly.

With a known post ID, one reply page takes one twitter.web.fetch_post_comments call. Each detail lookup and each further page is another execution. Sources: the free contract data:twitter.web.fetch_post_comments.

Keyword search · Social Data contract · Read saved results