Cut the Noise With an OpenAI dots Plugin for X Social Listening

X turns one keyword into hundreds of short posts a day, so an OpenAI dots plugin for X social listening lives or dies on its filters. OpenAI launched dots on 29 September 2026, and a dot will read whatever you point it at, all of it, every time the task runs. RedReplier scores each X post before the dot sees it, which gives you the filters. This guide is about using them.

The short version: connect RedReplier as a custom plugin, ask your dot for TWITTER mentions with count_mentions first, then list_mentions sorted by RELEVANCE with HIGH and VERY_HIGH only. Let RedReplier's email alerts handle speed and let the dot handle triage.

What does X listening need from a dot?

X listening needs scoring, deduplication across runs and a status on each post, because volume is the whole problem on X. A keyword that returns nine Reddit threads a day returns hundreds of X posts, most of them one line long and a good share of them the same thought from different accounts.

RedReplier supplies those three things. Each mention with the TWITTER source carries a relevanceScore and a status of NEW, APPROVED or REJECTED, so the dot reads a ranked shortlist instead of a timeline. Our guide to X social listening covers the same triage problem without an agent in the loop.

The plugin setup is the same as for Reddit and lives in the OpenAI dots plugin for Reddit monitoring guide: developer mode, https://mcp.redreplier.com/mcp at chatgpt.com/plugins, and an OAuth sign-in with your RedReplier account. OpenAI has not documented that dots use developer-mode plugins as of 29 September 2026, so prove it with the first read below.

Does a dot have an X plugin of its own?

OpenAI has not published which apps its plugin ecosystem includes, so as of 29 September 2026 nobody outside OpenAI can say whether X has a plugin. The launch post says dots connect to "over 4,000 apps" through plugins and gives no breakdown. Check ChatGPT's Plugins tab on your own account.

Any X plugin still hands your dot raw posts. Reading X through your own developer access means paying X for reads and writing the scoring yourself, and our X API pricing page tracks what those reads cost. RedReplier collects X mentions on its side, so the plugin reads them behind your RedReplier sign-in with no X developer account.

What do I ask my dot for on X?

An X request to your dot counts first, filters hard, and explains only the top two. The count decides whether today's X volume is a digest or a firehose before the dot pulls a single post.

Use the RedReplier plugin. Read only: never approve, reject,
change or delete anything.
1. Call count_mentions with sources TWITTER and statuses NEW,
   from 24 hours ago. Tell me the number.
2. Call list_mentions with sources TWITTER, statuses NEW,
   scoreBuckets HIGH and VERY_HIGH, sort RELEVANCE, limit 20,
   from 24 hours ago.
3. If step 1 returned more than 200, use VERY_HIGH only.
4. Call explain_mention on the top two.
Give me author, score, one line on what they said, and the link.

Name sort=RELEVANCE every time, because RedReplier's spec declares no default order for list_mentions. RECENT is the other value, and on X it returns the last few minutes rather than the day's best posts. The from filter works on ingestedAt, the time RedReplier found the post, so a fixed 24-hour window lines up with what RedReplier collected rather than with when the author posted.

Why does the same X post keep showing up in my digest?

The same X post keeps showing up because its status is still NEW, and a dot on ChatGPT Pro has no documented way to change it. OpenAI's August 2026 help article says "Pro users can connect MCPs with read/fetch permissions in developer mode", and update_mention_status is a write. The same article says "Full MCP is only available to Business and Enterprise/Edu."

That split changes how you run X listening on each plan.

On Business Premium, the dot's four Custom Rules settings decide how far it goes: take action without asking, take action when you say so or if pre-approved, ask before taking action, and hand off to you. Pick "ask before taking action" for update_mention_status. A rejected X post only comes back when you ask for statuses=REJECTED, so a misjudged one disappears quietly.

How fast can a dot catch an X thread?

A dot catches an X thread as fast as its recurring task runs, and OpenAI publishes no minimum interval for recurring tasks as of 29 September 2026. Event triggers exist, but OpenAI's docs limit them to cases "when a connected service supports event monitoring", and RedReplier's agent docs list tools and no events.

Speed and triage are different jobs, so split them. RedReplier monitors in real time on every plan and emails new matches every hour on Starter, every 30 minutes on Pro and every 15 minutes on Business. Those alerts are the fast channel. The dot's job is the morning and afternoon triage that ranks, explains and trims what the alerts already flagged.

Does reading hundreds of X posts eat my ChatGPT limits?

Conversations with your dot do not count toward ChatGPT usage limits, according to OpenAI's launch post. Tasks the dot starts in Codex or ChatGPT Work do count, and the plan includes "an allowance for deeper work, with extended limits for the first month". OpenAI publishes no rate limits or task caps for dots as of 29 September 2026.

So the reason to filter hard on X is your attention, not your ChatGPT meter. A 200-post summary of a broad keyword is still 200 posts you did not need, and the limit of 20 in the prompt above is the line between a digest and a transcript.

Can my dot reply to X posts for me?

Your dot cannot reply through RedReplier, because RedReplier's API has no tool or endpoint that posts to X. Each mention carries an aiReplySuggestion, which the spec describes as a "Draft reply for a human to review before posting." OpenAI's dots docs say the same from the agent side: "Asking your dot to draft replies doesn't give it permission to send them."

A dot's own browser reaches x.com, and OpenAI says dots are taught to seek authorization before sending a message. Set posting to "hand off to you" and write X replies yourself. A reply under your brand's handle is a public statement, and it deserves a human reading the thread first.

What breaks on X specifically?

Three things break on X that barely register on Reddit.

The first is an unset sort, covered above. The second is a broad keyword. A product name that doubles as a common word matches enormous X volume and drags the score distribution down, so even HIGH and above returns a long list. Fix the keyword in RedReplier before you tighten the dot's filters.

The third is memory. Every mention includes an author, which on X is a handle, and your dot creates its own memories from connected apps. OpenAI's help centre says you cannot view or edit individual dot memories, and deleting the dot is the only way to clear them. Ask for summaries and links, and leave full post text out of the digest.

How does the dots route compare with Muse and Grok on X?

The dots route is the only one where the agent's maker says conversations with it sit outside the plan's usage limits. The Meta Muse connector for X social listening builds a custom connector against the REST spec and draws on Muse's weekly token allowance. The Grok bot template for X social listening ships a skill that calls the same API with a redreplier_ token. RedReplier's social media monitoring page covers the scoring all three rely on.

Frequently Asked Questions

Does the dots plugin need X API access of its own?

The RedReplier plugin needs no X API access. RedReplier collects the X mentions, and your dot reads them through the plugin after you sign in with your RedReplier account. There is no X developer account and no per-read bill from X.

Can my dot listen to X and Bluesky in the same digest?

RedReplier's sources filter takes more than one value per call, so TWITTER and BLUESKY return together in one list_mentions response. Count each source separately first, because X volume buries Bluesky in a shared cap. The OpenAI dots plugin for social listening covers all five networks.

Which ChatGPT plans include dots?

Dots come with ChatGPT Pro, at $100, $200 or $500 a month, and with Business Premium. Enterprise, Edu and Healthcare get a beta that stays off until an admin turns it on. The first dot costs nothing extra on either plan.

Can I use a dot for X listening in the EU or UK?

Pro users in the European Economic Area, the UK and Switzerland cannot use dots as of 29 September 2026. Business Premium includes dots in every supported ChatGPT region. Users under 18 cannot use dots on any plan.

What do HIGH and VERY_HIGH mean on an X post?

HIGH and VERY_HIGH are the top two of RedReplier's five relevance buckets, above MEDIUM, LOW and VERY_LOW. The bucket comes from the mention's relevanceScore. explain_mention returns the reason behind the score, which separates a buying question from a joke that happens to use your keyword.

How do I stop my dot's X routine?

The Activity View shows the dot's current and delegated tasks, and OpenAI's launch post says you can pause or stop the dot's work. To cut RedReplier access, remove the plugin from the Plugins tab. Clearing the X handles the dot remembered takes deleting the dot.

How do I add the RedReplier plugin to my dot?

Turn on developer mode in ChatGPT, then add https://mcp.redreplier.com/mcp as a custom plugin at chatgpt.com/plugins. Sign in with your RedReplier account through OAuth. Run the count and list prompt once to confirm your dot can read it.

Can a dot on ChatGPT Pro mark X posts as rejected?

Not as far as OpenAI documents. OpenAI's August 2026 help article, which predates dots, gives Pro MCPs read/fetch permissions only, and update_mention_status is a write. The same article gives Business full MCP, so a dot on Business Premium can clear posts, ideally behind "ask before taking action".

Why should my dot call count_mentions before list_mentions?

The count tells you whether today's X volume is a digest or a firehose before the dot pulls a single post. If it returns more than 200, narrow the request to VERY_HIGH only. That keeps the list short enough to read.

Should I sort X mentions by RELEVANCE or RECENT?

Sort by RELEVANCE and name it in every request, because RedReplier's spec declares no default order for list_mentions. RECENT returns the last few minutes on X, not the day's best posts. Use it only when you want the newest arrivals.

Create your RedReplier account and give your dot a ranked X shortlist instead of a timeline, or read the agent docs first.

Sources: OpenAI's dots launch post, dots safety post, getting started help article, safety FAQ, dots user docs and the August 2026 MCP help article, plus the RedReplier OpenAPI spec, agent docs and pricing. Checked 29 September 2026.