BuzzyX

The X API reply restriction, explained for founders who reply by hand

Divv SaxenaPublished Last updated 9 min read

The X API reply rule since 23 February 2026: an API reply is allowed only if the original post mentions your account or quotes your post, otherwise it is rejected The short version of a rule that quietly reshaped reply tools.

Since 23 February 2026, X's self-serve API rejects a reply unless the original author @mentioned the replying account or quoted one of its posts. Enterprise is exempt. Replies you type on x.com or in the X app aren't API calls, so this rule doesn't apply.

BuzzyX is built on the X API, so this rule matters to me directly. I read X's own docs for it today, 6 November 2026, and I link every source below. Where I'm leaning on a developer's report instead of X's page, I say so.

What the X API reply restriction says

X put the change in its developer changelog on 23 February 2026, under the title "Addressing LLM-generated spam". The key line: programmatic replies via POST /2/tweets "are now only permitted when the original Post's author has 'summoned' the replier (by @mentioning that account or quoting one of its Posts)."

Before that day, any app could reply to any post by setting in_reply_to_tweet_id. After it, the app needs an invitation first. X's manage posts integration guide spells out what counts.

A quote card with X's wording from the manage posts integration guide: replies via the API are only permitted if the replying account has been explicitly summoned by the original post's author, by an @mention or a quote post X's wording, from the integration guide. I read it on 6 November 2026.

Two things count as a summon. The original author @mentions your account in their post, or the original author quotes one of your posts. That's the whole list. If neither is true, the guide says "the reply request will be rejected."

The same day, the @XDevelopers account posted the short version: you can only reply if the original author @mentions you or quotes your post, non-replies stay unchanged, and Enterprise and Public Utility apps aren't impacted. I'm linking Impress Watch's page because it embeds the post, and X's own developer forum thread, which the changelog points to, blocked my fetch on 6 November. So the changelog and the two docs pages are what I read first hand. The forum announcement loaded on 7 November, and I quote it below where it adds something.

Who it covers

The integration guide says the restrictions apply to "self-serve (Pay Per Use) API customers only." The @XDevelopers post names the older tiers too: Free, Basic, Pro and pay-per-use. Enterprise customers aren't subject to the limit.

A table of what the X API reply restriction covers: replies to posts that never mentioned you are rejected, replies to posts that mention or quote you are allowed, non-reply posts are unchanged, and Enterprise access is not impacted What's rejected, what's allowed and what never changed, with X's own words in the last column.

How the check works

The test runs on the post you are replying to, not on your app or the quality of your reply. Does that specific post @mention your account? Does it quote one of yours? If yes to either, the reply goes through. If no, it doesn't, however good the text is.

A flowchart: an app on a self-serve plan tries to reply, X checks whether the post mentions the account and whether it quotes one of the account's posts, and rejects the reply if neither is true One question, asked twice. Two noes and the reply is rejected.

Developers have reported what the failure looks like: a 403 with the message "You can only reply to or quote posts where you are mentioned or are the author." That wording comes from developer write-ups such as X-Autopilot's breakdown and TwitterAPI.io's 403 guide, not from the X pages I read, so treat the exact string as reported, not official. The same write-ups say replies to your own posts still work, which matters for threads. I haven't tested that myself.

What it doesn't change, and what else moved

Normal posts still go through. The summon check is only about replies, which is why scheduling your own posts through the API still works. One caveat from X's forum announcement: API posts may now only @mention or quote users who are already involved in the conversation, and that applies to normal posts as well as replies.

But replies weren't the only thing X tightened. The changelog says "additional restrictions apply to programmatically @mentioning or quoting users," and a pricing update effective 20 April 2026 removed Following, Likes and Quote-Posts via the API from all self-serve tiers. The write side of the API is a lot narrower than it was a year ago.

X's pricing page also shows how it sees replies. It lists "Post: Create (summoned)" at $0.010 per request, next to $0.015 for a normal post and $0.200 for a post with a URL.

Three price tiles from X's API pricing page: $0.010 for a summoned reply, $0.015 for a normal post and $0.200 for a post with a URL Per request, pay-per-use, read on 6 November 2026. Prices change, so check X's page before you build on them.

I read that as a signal. The invited reply gets the cheapest price. The reply to a stranger gets nothing, because it's blocked. I go deeper on these numbers in what the X API costs per call and the X API pricing guide.

What it means for reply tools and automation

Sort the reply tools you've seen into three groups, because the rule hits each one differently.

Tools that find posts by keyword and reply through the API. These are the "reply guy bots." On a self-serve plan, they can't reply to strangers anymore. The only route X points to is Enterprise, which is a different price bracket from a $29 tool.

Tools that answer people who mentioned your account. These still work. That's exactly the case the rule leaves open. A support bot that replies to mentions is fine by the API check.

Tools that draft the reply and let you post it yourself. The rule never touches them. The reply doesn't leave through the API at all.

There's also the policy layer, which was already there. X's automation rules said, when I checked them on 4 November, that "sending automated replies to posts based on keyword searches alone is not permitted." The API change is X enforcing in code what the rules page already said in words. My read: the crackdown was always coming, and the only surprise was how little room it left.

If a tool promises to reply to strangers for you, ask it two things. How does the reply reach X, through the API or some other route? And who presses send? If the answers are "not sure" and "the tool," I'd be worried about the account, because the account is what gets limited, not the tool.

Why BuzzyX's manual replies are unaffected

BuzzyX has never sent a reply through the API. I wrote down why BuzzyX never automates likes, follows, replies or DMs back in October, for quality and rules reasons. The API rule is a third reason that came free.

Here's how a reply works in the product. Discover shows you recent posts worth joining. On the paid plan you can ask for reply ideas in your voice and edit one. Then Reply on X opens X's own reply box with your text in it, and you press send on x.com. That reply goes out from your browser and your session, not from a POST /2/tweets call. The summon check never enters the picture.

Two paths side by side: a tool replying through the API, which X checks against the mention or quote rule, and replying from x.com with BuzzyX, where Reply on X opens X's own reply box and you press send The difference isn't the AI. It's where the reply leaves from and who sends it.

The free Chrome extension is even simpler. It reads what x.com already shows and ranks posts by freshness and room to reply. It doesn't post, like or follow anything.

The API calls BuzzyX does make are for things like publishing and scheduling posts you wrote. Those are not replies, so the summon check doesn't apply to them, though the @mention and quote limit above does.

I won't promise X never changes anything else. It already did, twice, in 2026. But a product whose rule is "you press the button" had the least to lose from this one.

FAQ

What is the X API reply restriction?

A rule X added on 23 February 2026. On self-serve API access (legacy Free, Basic and Pro, and pay-per-use), a reply sent through POST /2/tweets only goes through if the original post's author @mentioned the replying account or quoted one of its posts. Otherwise X rejects it. Enterprise access isn't affected.

Does it affect replies I type directly on X?

No. The rule is about replies sent through the X API. X's changelog, its integration guide and its forum announcement only describe API requests, and none of them mention replies you write and send yourself on x.com or in the X app.

Does BuzzyX send replies automatically?

No. BuzzyX has never sent a reply through the X API. Discover can suggest reply ideas in your voice, and Reply on X opens X's own reply box with your text in it, so you press send on x.com. The free Chrome extension ranks posts as you scroll and never posts, likes or follows.

Why did X change this?

X gave its reason in the developer forum announcement: programmatic replies were being used "to insert low-quality or automated content into conversations." Its changelog entry is titled "Addressing LLM-generated spam." X hasn't published more detail than that in the pages I read.

What should founders do instead?

Reply by hand on x.com, on fresh posts where you have something real to add. If you build on the API, reply to posts that mention or quote your account, which the rule still allows. For the finding part, see how to find tweets to reply to in 15 minutes a day.

What to do this week

A timeline of X's published changes: the reply rule on 23 February 2026, the pricing update on 20 April 2026 and the date these docs pages were re-read, 6 November 2026 The timeline from X's own pages.

If you use a reply tool, find out how it sends. If you build on the API, subscribe to mentions and reply to those. A reply to someone who tagged you is allowed, and it's the nicest kind of reply to write anyway. If you reply by hand, keep going. Finding the right posts is the slow part, and I covered it in how to find tweets to reply to in 15 minutes a day. For what to actually write, see the guide to growing on X with replies.

Want the finding sped up while you stay in control? Add the free BuzzyX extension to Chrome. Reply ideas in your voice, scheduling and analytics come with the BuzzyX plan at $29 a month, and you can try it free for 7 days: start your 7-day trial.

BuzzyX · X growth workspace

Grow on X with posts that sound like you.

Compose in your voice, schedule the week, find posts worth replying to and read real analytics from the X API. $29/mo after the trial, everything included.

Try BuzzyX free →See pricing7-day free trial. Card required, $0 today, cancel in Settings.

Share

All posts →

All posts