Post, delete, search and read tweets, look up users, and like/retweet on X (formerly Twitter).
Twitter (X) ships in the w6w first-party pack. It declares 8 actions, 2 health checks, and the host runs its code in a sandbox that never sees the credential.
io.w6w.twitterThis app posts to X directly from a workflow — plain text, or a tweet with an image or GIF attached — deletes a tweet, and reads back a specific one by id. Search reaches tweets from the last seven days, and a user can be looked up by username or read by their own recent tweet history.
Liking and retweeting are both supported and resolve to the authenticated user’s own account automatically, matching how X’s API itself scopes those actions. Authentication runs through OAuth with PKCE, X’s only supported flow for acting as a user, with a refresh token requested up front so a connection doesn’t need reauthorizing every two hours.
Good for posting scheduled or triggered content, monitoring recent mentions or a search term, and pulling a user’s tweet history into another system. Since 2026, most X API usage bills per request rather than through a flat free tier, worth knowing before wiring this into anything high-volume.
Three routes to the same 8 actions. The Workflow tab is generated from Twitter (X)'s own manifest and carries its real ids, so it is copy-pasteable; the Code and CLI examples are the same call for any action on any app, so every app-specific value in them is a blank you fill in.
tweet-create Post a tweet, optionally with one image/GIF attached, a reply, or a quote tweet.
A workflow step names the app and the action, and the editor fills in the
connection when you pick one. This is the Step shape from the
workflow spec, carrying Twitter (X)'s real ids.
{
"manifestVersion": "2",
"name": "twitter-example",
"steps": [
{
"id": "tweet-get",
"uses": {
"app": "io.w6w.twitter",
"action": "tweet-get",
"connection": "conn_YOUR_CONNECTION_ID"
},
"with": {
"tweetId": "<tweetId>"
}
}
]
}tweet-get tweet-create tweet-like tweet-retweet tweet-search-recent +3 more actions available
Every app-specific value here is a blank you have to fill in. An
app action is reached through the connection that authenticates it, so the
address is a connection id, not the app id — and connections belong to your account,
so a public page cannot know yours. Create one for Twitter (X), then fill in
the three blanks: conn_YOUR_CONNECTION_ID, the action key, and the
parameters that action declares. The call itself is real — the shape is transcribed
from the studio's own snippet builder, which prints the same kind of blanks — but
nothing in it is specific to Twitter (X). The Workflow tab is where this app's
real ids are.
npm install @w6w/sdkyarn add @w6w/sdkpnpm add @w6w/sdkdeno add npm:@w6w/sdkimport { W6wClient, isActionRun } from "@w6w/sdk";
// Reads W6W_BASE_URL and W6W_TOKEN from the environment when omitted.
const client = new W6wClient();
const envelope = await client.run({
urn: "conn_YOUR_CONNECTION_ID",
action: "tweet-create",
payload: {
// text: "<value>",
// media: "<value>",
// mediaMimeType: "<value>",
// mediaCategory: "<value>",
// replyToTweetId: "<value>",
// quoteTweetId: "<value>",
},
});
if (isActionRun(envelope)) console.log(envelope.value); npm install -g @w6w/cli w6w run conn_YOUR_CONNECTION_ID --action tweet-create --payload '{}' Give an AI agent Twitter (X) — without giving it Twitter (X)'s credentials. One MCP endpoint exposes every app, function and workflow the caller is entitled to, as tools it can discover and run. Access is granted per team while we onboard.
One tool call{
"name": "w6w_invoke",
"arguments": {
"ref": "app:io.w6w.twitter#tweet-get",
"input": {
"tweetId": "<tweetId>"
}
}
}
Every tool names its target with a single ref. The
app: form above doesn't name a connection at all — the
host resolves which of the caller's Twitter (X) connections to sign
with, and refuses rather than guesses when the answer is ambiguous.
The token is attached host-side, at the moment of the call. It is never a tool argument, never in the model's context, and never in a transcript — so a prompt injection has nothing to exfiltrate.
Tools are derived per end user from what that person has actually connected and is entitled to — not one shared bot identity carrying the union of everyone's access.
Multi-step work runs on the workflow engine and returns a run handle the agent can poll — retries, branching and state survive the conversation that started them.
Twitter (X)'s declared health checks are on the surface too, so an agent can tell "the vendor is down" from "your credential expired" before it burns a retry on either.
The MCP surface is part of the hosted platform. Twitter (X) itself is MIT, and the runtime that executes it is source-available (FSL).
Twitter (X) declares its own checks, so its health is a property of the app rather than something the host guesses at.
No machine-readable status feed is documented for the X Developer Platform (developer.x.com/status is a human-readable dashboard only), so this app declares the absence rather than guessing at an undocumented endpoint.
Reads x-rate-limit-* off GET /2/users/me. Reports only that endpoint's 15-minute window — X exposes no aggregate quota endpoint, and this does not reflect the pay-per-use dollar-credit balance (see README).