tweetkit-x

Post, delete & read on X (Twitter) — from Python and as an MCP server — using your own logged-in browser session.
No paid API. No developer app. No OAuth dance. Just your session cookie.

Quick start · Get your cookie · Python · CLI · MCP server · How it works · Disclaimer

Why

Since February 6, 2026, X's official API has no free tier for new developers — it's pay-per-use: ~$0.015 per post, ~$0.20 per post with a link, plus per-read charges. For publishing your own tweets (a bot posting your content, a scheduled thread, a cleanup script that deletes old posts), that's a recurring bill for something your browser already does for free.

tweetkit-x skips the API entirely. It replays the exact internal HTTP calls your browser makes when you click Post, Delete, or scroll your profile — authenticated with the session cookie you already have. Post, delete, and read the full timeline, at zero API cost.

⚠️ Automating your web session is against X's Terms of Service. This is a grey-area tool for personal automation. Account risk is entirely yours. See the Disclaimer.

Features

• ✅ Post — text, up to 4 images, GIF & video (with alt text), replies, threads, quote-tweets, long-form note tweets, and scheduled tweets
• 🗑️ Delete — one id, a list of ids, or "find matching & delete" cleanup loops
• ❤️ Engage — like, retweet, bookmark, follow, block, mute, pin — each with an undo
• 🌐 Search all of X — the real SearchTimeline (operators like from:, min_faves:, filter:media), plus local regex over your own tweets
• 📖 Read — your home feed, any timeline, a tweet's replies/conversation, followers/following, likers/retweeters, bookmarks, a user's likes, notifications, and full profiles
• 🔌 MCP server — 40 tools; drop into Claude Desktop / Claude Code / Cursor; tools that explain the cookie and set it up with one paste
• 🔑 Frictionless auth — paste a Cookie: header, drop a HAR file, or a storage-dump zip; store in a file or the macOS Keychain
• 🐍 Python API and a CLI (tweetkit …)
• 🪶 Tiny footprint — no headless browser, no Selenium

Quick start

pip install "tweetkit-x[mcp]"      # or: pip install tweetkit-x   (library + CLI only)

# 1. Import your session cookie (three ways — see the next section)
tweetkit import --paste                       # paste the Cookie: header, Ctrl-D
#   or:  tweetkit import --file x.com.har
#   or:  tweetkit import --file storagedump_x.com.zip --keychain tweetkit

# 2. Verify (no network call)
tweetkit whoami
# {"ready": true, "user_id": "1986...", "ct0_len": 160}

# 3. Use it
tweetkit post "hello world 👋"
tweetkit search "xmr|monero" --regex          # your tweets that match
tweetkit delete 2063240404980445603

Getting your session cookie

The only secret tweetkit needs is your X session cookie — the same Cookie: header your logged-in browser sends. It must contain at least two values:

❗ auth_token is HttpOnly, so a document.cookie copy in the JS console will NOT include it. Use one of the reliable methods below.

Method A — copy the Cookie: header (most reliable, one paste)

1. Open x.com while logged in.
2. Open DevTools (F12) → the Network tab.
3. Click any request to x.com (e.g. HomeTimeline).
4. Under Request Headers, find cookie: and copy the entire value.
5. Feed it in:

tweetkit import --paste
# paste the cookie header, then press Ctrl-D

Method B — an F12 "Preserve log" HAR file

1. Open x.com logged in → DevTools (F12) → Network.
2. Tick "Preserve log", then reload / click around so requests are captured.
3. Right-click any request → "Save all as HAR" (or use the ⬇ export icon).
4. Import it:

tweetkit import --file ~/Downloads/x.com.har

Modern Chrome can redact cookies from exported HARs. If tweetkit reports auth_token missing, either enable "Allow to generate HAR with sensitive data" in the Network settings, or fall back to Method A / Method C.

Method C — the storagedump browser extension (zip)

The easiest no-DevTools route. storagedump (a Chrome extension built by this project's author) exports a tab's browser storage — including the HttpOnly auth_token, which document.cookie can't reach — as a .zip containing cookies.json. On x.com (logged in), click the extension → export, then:

tweetkit import --file storagedump_x.com_2026-07-05.zip

tweetkit reads the extension's native format, { "data": [ { "key": "...", "value": "..." }, ... ] }, and also accepts a plain list or a {name: value} map. Cookie-Editor / EditThisCookie JSON exports (--file export.json) work too.

Install: storagedump on the Chrome Web Store.

Where the cookie is stored

By default tweetkit import writes ~/.config/tweetkit/cookie.txt (chmod 600; also git-ignored). Prefer the macOS Keychain? Add --keychain <slug>:

tweetkit import --paste --keychain tweetkit
export TWEETKIT_COOKIE_KEYCHAIN=tweetkit

Resolution order at runtime (first hit wins): explicit arg → TWEETKIT_COOKIE → TWEETKIT_COOKIE_FILE → ./cookie.txt → ~/.config/tweetkit/cookie.txt → Keychain (TWEETKIT_COOKIE_KEYCHAIN).

Python usage

from tweetkit_x import TweetKit

tk = TweetKit(keychain_slug="tweetkit")          # or cookie="auth_token=...; ct0=..."

tk.whoami()                                       # {'ready': True, 'user_id': '...', 'ct0_len': 160}

# post
tk.post("hello world")
tk.post("with a picture", image_path="chart.png")
tk.post("a reply", reply_to="1800000000000000000")

# thread — strings or (text, image_path) tuples; each replies to the previous
tk.post_thread(["1/ intro", ("2/ chart", "chart.png"), "3/ outro"])

# read
mine = tk.get_tweets(limit=100)                   # your tweets, newest first
theirs = tk.get_tweets(username="jack", limit=50) # anyone's

# search (local filter over the timeline — no paid search API)
hits = tk.search("monero")                        # substring
hits = tk.search(r"xmr|monero", regex=True)       # regex

# delete
tk.delete("1800000000000000000")
tk.delete_many([t["id"] for t in hits])

# quote-tweet
tk.quote("this is great", quote_tweet_id="1800000000000000000")

# engage (each has an undo)
tk.like("1800000000000000000");     tk.unlike("1800000000000000000")
tk.retweet("1800000000000000000");  tk.unretweet("1800000000000000000")
tk.bookmark("1800000000000000000"); tk.unbookmark("1800000000000000000")

# read a single tweet, and search ALL of X (real search + operators)
tk.get_tweet("1800000000000000000")
tk.search_x("mcp server", latest=True, limit=40)
tk.search_x("from:jack min_faves:1000")     # X search operators work

Every write returns {'ok': True, 'id': '...', 'url': '...'} on success, or {'ok': False, 'status': ..., 'error': '...'} on failure — easy to log or retry.

See examples/ for a full "find my $XMR tweets and delete them" cleanup loop.

CLI usage

tweetkit whoami
tweetkit post "just some text"
tweetkit post "text + media" --image pic.png
tweetkit post "a reply" --reply-to 1800000000000000000
tweetkit thread thread.txt                 # tweets separated by blank lines
tweetkit delete 1800000000000000000 1800000000000000001
tweetkit tweets --limit 100                # your tweets
tweetkit tweets @jack --limit 50           # someone else's
tweetkit search monero                     # your tweets containing "monero"
tweetkit search "xmr|monero" --regex
tweetkit searchx "mcp server" --latest     # search ALL of X (real search + operators)
tweetkit get 1800000000000000000           # one tweet by id
tweetkit quote "great thread" 1800000000000000000
tweetkit like 1800000000000000000          # also: unlike / retweet / unretweet / bookmark / unbookmark

A thread.txt is just tweets separated by blank lines.

MCP server