I would like to recommend building a conversations API that enables developers to track updates to full conversation threads. This API would retrieve the entire conversation that a Tweet is in, and return all Tweets and replies within a given thread. This API would enable developers to go from a Tweet of interest to all replies in the thread, without having to recursively search for Tweets with the correct "inreplytostatusid".43 votes
It would be really helpful to have an analytics endpoint in which the entity was a media ID (in my case a video ID) and you could request engagement & video metrics aggregated at the video level, even if that video appeared across many different tweets. Right now this is possible through the Engagement API for video views only, but it would be great if it were available through the Ads API & included seconds/minutes viewed, percentile completions, call to action clicks, etc.35 votes
It would be useful to be able to retrieve more than 100 results calling GET statuses/retweeters/ids
The current limitation doesn't make sense.18 votes
We need a way to manage muted keywords, similar to the existing endpoints for muted users:
POST mutes/keywords/destroy12 votes
Why don't we have the same kind of statuses/retweeters/ids endpoint but for likes, in order to get all the user ids who liked a particular tweet ?11 votes
Thanks for the request & details. We don’t yet have plans to address this, but I have shared your request with the Product team for future consideration.
When I try to hide a Tweet in a thread that wasn’t started by the authenticating user, I get the response:
[title] => Forbidden
[type] => about:blank
[status] => 403
[detail] => Forbidden
I understand that this action should actually be forbidden, but the error in the API response doesn't tell me why the action is forbidden (which is that the authenticating user is trying to hide a Tweet in a thread that wasn't started by the authenticating user)11 votes
Add a field to know when media is being withheld by country9 votes
The standard search API is somehow regarding screen name as part of text body and therefore is also searching in screen names. That can cause the search results to contain irrelevant tweets. For example, searching the word "nature" can return @human_nature's tweets whose text bodies don't contain the word "nature".
A similar issue applies to user mentions. For example, searching the word "nature" can return tweets which contain user mention @human_nature and don't contain the word "nature".
Please consider implementing an option which prevents the standard search API from searching in screen names and user mentions. Thanks.8 votes
Developers may need to retrieve a list of hidden replies for a specific Tweet. This can be useful to obtain a list of Tweets can can be unhidden, given appropriate user consent. Developers can also create aggregate reporting via this API (for example, count how many Tweets are hidden per conversation, or per time period).7 votes
Include message lifetime reach (unique impressions) and profile-level daily reach metrics for messages. This provides better insight into the number of individuals that saw a specific tweet or content associated with a profile/brand.7 votes
Is there any plan to allow access to the cropping rectangles that are associated with each media image? These are called focusrects and are nested under the originalinfo property of extended_entities.media and could be passed along with the rest of the media info in the API request.7 votes
Really often, people on Twitter define a choice on their poll just to let people not concerned see the result. Why not providing a simple button to let anybody see the result without having to choose an answer ?6 votes
Add an API version of the popular "Who to follow" link seen at lower right on the home page: https://twitter.com/<...>/whotofollow .
(Currently it uses non-public APIs under the hood.)6 votes
I was trying to summarize a conference in a moment with over 64 tweets. Trying to organize those manually moving them up and down 1-by-1 on the site was a pain (as many people tweet things not quite in chronological order of the conference or if you sorted once and want to add more later, it's hard).
It'd be great to expose the API for creating/editing moments so that other tools could help push the experience of capturing events or help to correlate topics and insert into proper places.6 votes
It's useful to have the date of last activity of a user. Before we could deduce it from the most recent tweet but now it's gone in v2 and it become really tricky to get this information.4 votes
Support for the sample: operator. Many use cases would benefit from random samples. For example, sample:10 would return 10% across each hour...4 votes
While I understand the motivation for the separate "expansions" in the GET Tweets endpoint, I wonder if you have considered using GraphQL to allow the user to determine the shape of the tweet data returned?
In adding support for the labs endpoint to twarc  I found that I wanted to knit back together the data that is separated out into tweets. If GraphQL was used I don't think this would be necessary.
I would be curious to know if Twitter Labs considered switching to GraphQL for fetching tweets and users.4 votes
To enable one to send a tweet to a group4 votes
Build a new request operator that can be included in an API request, and returns a translation of the Tweet text in the Tweet payload. Supported languages should be the same languages that are supported by the "lang:" request operator.4 votes
Wildcards would be very nice to have in filtered streams: Customers want them, and there is no way to emulate it on our side.
When a query term contains a
*would match any character in a tweet, one or more time. When a query term contains a
?would match any character, exactly one time.
- Don't see your idea?