X-use

Community

Browser-native AI agents for X (Twitter): multi-account, MCP-ready, no X API key required.

How to install ↓ Compare GitHub source repository ↗

Category: Browser #13 of 32
Stars: ★ 177 Very popular
Transport: stdio (local process)
Runtime: Python 3.10+ · Prebuilt binary
Credentials: Optional API key
License: MIT
Last commit: 8 days ago
Tools: 33

55FMRS · C

x-use is an open-source, MIT-licensed MCP server that automates X (Twitter) without an API key via browser automation. It provides 33 MCP tools, supports multi-account, draft approval, proxy management, and integrates with clients like Claude Desktop. Main risk is X account suspension, suitable for individuals and small teams.

Strongest · Documentation 13/20 Weakest · Reliability 8/20

Reliability

8/20

Security and permissions

10/20

Maintenance

12/20

Documentation

13/20

Setup experience

12/20

Why each score

Reliability 8/20: Evidence: The project includes 599 tests and a CI workflow (Python 3.10/3.11/3.12) covering key server paths like MCP tool contract, sessions, and queue. Tests require no network or browser, adding confidence. Deductions: Nothing was executed during static review, so actual runtime errors and behavior are unverified. Tests may pass under mocks but real browser automation could fail. The cap is 12, hence a relatively lower score of 8.
Security and permissions 10/20: Evidence: The server handles sensitive cookies and proxy credentials but masks them in resources and strips secrets from account views. Write tools are draft-gated by default, requiring explicit approve_draft, preventing unconfirmed publishing. Environment variables are marked as secret (OPENAI_API_KEY) and disclosure is explicit. Deductions: While security boundaries exist (e.g., data storage, external network), the full least-privilege and security scoping cannot be fully confirmed from static inspection. No red-line violations found.
Maintenance 12/20: Evidence: The project has an MIT license, CI badges shown in README, and a test suite of 599 tests. Continuous development is indicated (roadmap, contributing guide). Deductions: No clear release cadence or maintainer response times; dependency updates not explicitly visible; versioning exists but details are limited. Hence lower than full marks.
Documentation 13/20: Evidence: Extensive README covering installation, configuration, MCP tools, configuration reference, best practices, and FAQ. Includes an MCP guide with examples. Includes setup prompts and skills. Deductions: While thorough, some specifics on limitations (rate limits, concurrency) might be missing, and troubleshooting clarity is not fully explicit. Documentation consistency with actual functionality is unverified.
Setup experience 12/20: Evidence: Clear installation instructions including PyPI install, one-liner installer, and manual steps. Client-specific config examples are given (claude_desktop_config.json). Includes init wizard and doctor command. Deductions: Since static review, install scripts not executed; external dependencies (Chrome, proxies possibly) may make it fragile. Lacks explicit step-by-step troubleshooting. Hence below maximum.

Static review · not runListed 2026-08-07

Read the FMRS scoring method →

Fit and risk

What it can accessUses the networkChanges third-party account dataControls a browser

Best for

Individuals or small teams who want to automate X accounts without paying X API fees
Developers using MCP clients like Claude Desktop, Cursor
Users managing multiple X accounts

Not for

Enterprises requiring full official X API features (analytics, ads)
Users concerned about X account suspension due to browser automation risks
Users needing production-scale deployment, as Docker is not yet available

Required permissions

Requires X account cookies for browser session
Requires Chrome browser
Optional: LLM API key (only for server-side text generation)
Optional: Proxy (recommended for multi-account)

Risks and side effects

X accounts may be suspended for violating terms of service
Browser automation may be detected by X, leading to bans
Cookies and credentials may be exposed despite masking
Draft mode is on by default, but if turned off, publish actions will execute directly

Setup

Before you start

Runtime：Python 3.10+ · Prebuilt binary

OPENAI_API_KEY optionalsecret Optional key, only needed for server-side text generation (auto replies, composite draft tools, background automation); interactive MCP use is keyless.

Other optional settings (1)

OPENAI_BASE_URL optional Optional OpenAI-compatible endpoint, e.g. OpenRouter or local Ollama (http://localhost:11434/v1); defaults to api.openai.com.

Install with pip install x-use-mcp.
Run x-use init for interactive setup wizard.
Run x-use doctor to verify environment.
Add {"mcpServers": {"x-use": {"command": "x-use", "args": ["mcp"]}}} to your MCP client config (e.g., Claude Desktop).
Restart the client.

claude_desktop_config.json

{
  "mcpServers": {
    "x-use": {
      "command": "x-use",
      "args": [
        "mcp"
      ]
    }
  }
}

Shown for Claude Desktop. Other clients may use a different file or key (VS Code uses "servers") — the configurator below converts it.

.vscode/mcp.json

{
  "servers": {
    "x-use": {
      "command": "x-use",
      "args": [
        "mcp"
      ]
    }
  }
}

Goes in your project's .vscode/mcp.json (VS Code uses a "servers" key).

Terminal

claude mcp add x-use -- x-use mcp

Run it in a terminal; replace any <…> placeholders with your own values first.

Generate and validate client config

Check that it works

Restart the client and call list_accounts; a returned (possibly empty) account list confirms the MCP server is connected. You can also run x-use doctor to verify the environment.

Troubleshooting

Run x-use doctor to check environment
Ensure Chrome is installed and compatible
Check draft_mode setting in config/settings.json
Ensure MCP client can find x-use command (PATH)
Check logs/accounts/<account_id>.jsonl logs

Things to try

Once connected, you can ask your AI assistant things like:

List all my configured X accounts and show their health status
Search recent tweets about AI agents and prepare a reply draft for one of them
Draft three post variations about local AI models without publishing anything
Show what's currently in the scheduled action queue

Tools 33

list_accounts read-only

List all configured accounts.

get_account read-only

Get a single account's full masked configuration.

get_metrics read-only

Get metrics (e.g., counters) for a specified account.

search_tweets read-only

Search tweets using keywords.

search_profile read-only

Search user profiles.

get_tweet read-only

Get tweet details, including images.

prepare_reply read-only

Prepare a reply draft for a tweet (requires approval).

list_queue read-only

List the scheduled action queue.

Show 25 more tools

list_drafts read-only

List drafts pending approval.

get_draft read-only

Get draft details.

reject_draft writes

Reject a draft.

get_run_status read-only

Get background run status.

get_account_health read-only

Check account health status.

list_proxies read-only

List configured proxy pools.

post_tweet writes

Post a tweet (draft by default).

generate_and_post writes

Generate and post a tweet (requires LLM key, draft by default).

reply_to_tweet writes

Reply to a tweet (draft by default).

engage writes

Perform engagement actions like likes, retweets (draft by default).

run_cycle writes

Run an automation cycle (e.g., keyword replies).

approve_draft writes

Approve a draft and actually execute the publish/engage action.

queue_post writes

Queue a post for later.

queue_engagement writes

Queue an engagement action.

cancel_queued_action writes

Cancel a queued action.

process_queue writes

Process the queue of pending actions.

research_and_stage writes

Use server-side LLM to research and stage drafts.

draft_post_variations writes

Use server-side LLM to generate draft post variations.

add_account writes

Add a new account.

update_account writes

Update account configuration.

set_account_active writes

Activate or deactivate an account.

remove_account destructive

Remove an account.

add_proxy writes

Add a proxy.

remove_proxy destructive

Remove a proxy.

test_proxy read-only

Test proxy connectivity.

Use cases

Automate posting and replying on X

Manage multiple accounts with separate configurations

Keyword search, targeted engagement, and content curation

Control X accounts directly from MCP clients like Claude Desktop

Supported clients

Claude Desktop

Listed from the project's documentation, not tested by this site.

Overview

x-use drives a real, stealth-hardened browser instead of the paid X API. It posts, replies, searches, and engages across as many accounts as you configure. It writes content with your own LLM and exposes everything as MCP tools, so Claude Desktop, Claude Code, Cursor, and other MCP clients can run your X presence directly. By default, write actions go through a draft-approval step, so an agent can prepare work all day while nothing reaches X until a human says yes.

Similar servers

Playwright MCP 78 · B

Microsoft's official Playwright-based browser automation MCP server

★ 38.3k · Tools 10 Compare with this →

Chrome DevTools MCP 76 · B

The Chrome team's official MCP server exposing Chrome DevTools capabilities to AI assistants

★ 53.8k · Tools 22 Compare with this →

FSB MCP Server 66 · C

Control the FSB browser extension from any MCP client for DOM-first AI browser automation.

★ 195 · Tools 30 Compare with this →

Playwright MCP Server 66 · C

Playwright-based browser automation MCP server

★ 5.7k · Tools 1 Compare with this →

Compare with other MCP servers

Source revision b3df6412a7d3 Data synced 2026-10-19 Read the FMRS scoring method