x-mcp-playwright
A Twitter/X MCP server powered by patchright (stealth Playwright fork) and a persistent Chrome user profile. It enables posting, engaging, searching, and managing tweets, threads, media, and user interactions via natural language.
README
x-mcp-playwright
A Twitter/X MCP server powered by patchright (stealth Playwright fork) and a persistent Chrome user profile.
9 grouped tools — 33 actions preserved, zero functionality lost.
1. Install
git clone <this-repo> x-mcp-playwright
cd x-mcp-playwright
python -m venv .venv
# Windows:
.venv\Scripts\activate
# macOS/Linux:
source .venv/bin/activate
pip install -e .
patchright install chromium
One-time login
x-mcp-login
# or: python login.py
Opens a real browser window. Log in to X normally, then close the window. Session saved in ~/.x-mcp-playwright/profile/.
2. Register in your MCP client
Claude Code
claude mcp add --scope user x-twitter x-mcp-server
Claude Desktop
{
"mcpServers": {
"x-twitter": {
"command": "x-mcp-server",
"env": { "X_MCP_HEADLESS": "1" }
}
}
}
If x-mcp-server isn't on PATH, use:
{
"mcpServers": {
"x-twitter": {
"command": "/absolute/path/to/.venv/bin/python",
"args": ["-m", "x_mcp_server"]
}
}
}
Docker (headless VPS)
docker build -t x-mcp-playwright .
docker run -i --rm \
-v ~/.x-mcp-playwright/profile:/root/.x-mcp-playwright/profile \
x-mcp-playwright
3. Tool catalog (9 tools, 33 actions)
x_post — Publish tweets (6 actions)
| action | Required args | Description |
|---|---|---|
post |
text |
Post a single tweet |
thread |
tweets (newline-separated) |
Post a connected thread |
post_media |
text, media_paths (comma-separated) |
Up to 4 images or 1 video |
reply |
tweet_url, reply_text |
Reply to a tweet |
quote |
tweet_url, comment |
Quote-tweet with comment |
delete |
tweet_url |
Delete your tweet |
x_engage — Interact with tweets (4 actions)
| action | Required args | on param |
|---|---|---|
like |
tweet_url |
on=True to like, False to unlike |
retweet |
tweet_url |
on=True to RT, False to undo |
bookmark |
tweet_url |
on=True to save, False to remove |
pin |
tweet_url |
on=True to pin, False to unpin |
x_user — User operations (6 actions)
| action | Required args | Optional |
|---|---|---|
profile |
handle |
— |
timeline |
handle |
tab="tweets"|"likes"|"media", limit |
connections |
handle |
kind="followers"|"following", limit |
follow |
handle |
on=True to follow, False to unfollow |
block |
handle |
— |
mute |
handle |
— |
x_search — Search tweets (simple + advanced)
| arg | Default | Description |
|---|---|---|
query |
— | Keywords to search |
filter |
"top" |
"latest", "people", "media" |
from_user |
— | Filter by author |
to_user |
— | Filter by reply-to |
since / until |
— | Date range (YYYY-MM-DD) |
min_likes |
0 | Minimum likes |
min_retweets |
0 | Minimum retweets |
lang |
— | Language code |
limit |
10 | Max results |
x_feed — Browse feeds (4 actions)
| action | Description |
|---|---|
home |
Your home timeline. Optional: tab="for_you"|"following" |
bookmarks |
Your saved bookmarks |
notifications |
Mentions, likes, follows |
trending |
What's happening |
x_tweet — Tweet details (4 actions)
| action | Required args | Description |
|---|---|---|
details |
tweet_url |
Full tweet with metrics |
replies |
tweet_url |
Replies under a tweet |
analytics |
tweet_url |
Impressions/engagements (own tweets) |
engagers |
tweet_url |
Who liked/retweeted (kind="likes"|"retweets") |
x_list — Lists
| Scenario | Args |
|---|---|
| Your lists | No args |
| User's lists | handle |
| List tweets | list_id |
x_dm — Direct messages (3 actions)
| action | Required args | Description |
|---|---|---|
send |
handle, message |
Send a DM |
conversations |
— | List recent DM threads |
messages |
handle |
Read messages with a user |
x_session — Session management (2 actions)
| action | Description |
|---|---|
check |
Returns {logged_in, current_url, user_handle} |
screenshot |
Capture any X page. Optional: url, full_page |
4. Environment variables
| Variable | Default | Description |
|---|---|---|
X_MCP_HEADLESS |
1 |
0 to see the browser window |
X_MCP_TIMEOUT_MS |
20000 |
Default Playwright timeout (ms) |
X_MCP_SCREENSHOT_DIR |
~/.x-mcp-playwright/screenshots |
Debug screenshots |
X_MCP_USE_SYSTEM_CHROME |
0 |
1 to use system Chrome |
5. Safety
- Rate limiter: Write actions (post, delete, DM, reply) capped at 1/min. Engage actions (like, RT, follow) capped at 2/min. Read actions unlimited.
- Session keepalive: Background task pings X every 30 min. Logs warning if session expires.
- Pinned to your real account. Anything the LLM does, you did.
6. MCP resources
x://session/status— Login statex://version— Server versionx://tools/list— All toolsx://tools/categories— Grouped taxonomy
7. MCP prompts
analyze_tweet(tweet_url)— Fetch & analyze a tweet (sentiment, engagement, response tone)draft_reply(tweet_url, intent)— Draft 3 reply variants under 280 chars
7. Development
make install # deps + chromium
make login # interactive X login
make test # unit tests (no browser)
make smoke # live search test
make server # run MCP server
make lint # compile-check
make clean # remove pycache, screenshots
8. Safety
- Pinned to your real account. Anything the LLM does, you did.
- Cap: <= 1-2 publish/reply actions per minute.
- Gate destructive tools (
delete,block,send_dm) behind human approval on VPS.
License
MIT
Recommended Servers
playwright-mcp
A Model Context Protocol server that enables LLMs to interact with web pages through structured accessibility snapshots without requiring vision models or screenshots.
Audiense Insights MCP Server
Enables interaction with Audiense Insights accounts via the Model Context Protocol, facilitating the extraction and analysis of marketing insights and audience data including demographics, behavior, and influencer engagement.
Magic Component Platform (MCP)
An AI-powered tool that generates modern UI components from natural language descriptions, integrating with popular IDEs to streamline UI development workflow.
VeyraX MCP
Single MCP tool to connect all your favorite tools: Gmail, Calendar and 40 more.
graphlit-mcp-server
The Model Context Protocol (MCP) Server enables integration between MCP clients and the Graphlit service. Ingest anything from Slack to Gmail to podcast feeds, in addition to web crawling, into a Graphlit project - and then retrieve relevant contents from the MCP client.
Kagi MCP Server
An MCP server that integrates Kagi search capabilities with Claude AI, enabling Claude to perform real-time web searches when answering questions that require up-to-date information.
E2B
Using MCP to run code via e2b.
Neon Database
MCP server for interacting with Neon Management API and databases
Exa Search
A Model Context Protocol (MCP) server lets AI assistants like Claude use the Exa AI Search API for web searches. This setup allows AI models to get real-time web information in a safe and controlled way.
Qdrant Server
This repository is an example of how to create a MCP server for Qdrant, a vector search engine.