Upload and schedule TikTok videos from the command line or Python, with a proxy per account.

uv tool install autotok · one login · one command per upload

_{Click the video to watch it with sound.}

uv tool install "autotok[youtube]"
autotok install-browser                                   # one-time browser download, about 300 MB

Then log in once per account, and upload with one command:

autotok login -n my_account                               # log in on the page that opens; the QR code is quickest
autotok upload -u my_account -v clip.mp4 -t "Hello #fyp"

No uv? pipx install "autotok[youtube]" or pip install "autotok[youtube]" work too. Getting started goes through each step, including proxies and cloud browsers.

Swiftproxy provides 90M+ clean residential and static residential proxies designed for social media automation, multi-account management, and web automation workflows. With stable connections, global IP coverage, and reliable proxy infrastructure, Swiftproxy helps users manage multiple accounts securely, reduce IP-related restrictions, and scale their automation tasks. Residential proxy traffic never expires until used, and free testing is available.

Get 10% off with code PROXY90 — Get proxies from Swiftproxy →

Termius provides a secure, reliable, and collaborative SSH client.

Get Termius →

Setup takes about five minutes. After installing (step 1), you can run autotok setup to be walked through the rest: it checks the browser, asks for a proxy and tests it, then logs you in to TikTok through it. The steps below do the same by hand.

uv tool install "autotok[youtube]"

That's it: autotok is now a command you can run from any terminal (check with autotok --version). uv keeps autotok separate from your other Python packages, and downloads Python for you if you don't have it.

Don't have uv? Install it with one line from docs.astral.sh/uv, or use pipx: pipx install "autotok[youtube]".

• Update later: uv tool upgrade autotok.
• Try it without installing: uvx --from "autotok[youtube]" autotok --version.
• Using autotok from Python code? Add it to your project: uv add "autotok[youtube]" or pip install "autotok[youtube]".
• [youtube] adds YouTube-link support. Leave it out if you don't need it.
• autotok "not recognized" or "command not found"? Run uv tool update-shell once and open a new terminal.

autotok uses a browser for the one-time login and to sign each upload. Pick one:

For the local browser on a fresh Linux server, run autotok install-browser --with-deps as root, which also installs Chromium's system libraries.

autotok login -n my_account

A browser window opens on TikTok's login page. Log in however you like; the QR code with the TikTok app on your phone is quickest. The window closes by itself and the session is saved as my_account. With a cloud browser you get a link to open instead.

autotok accounts check my_account     # confirms TikTok accepts the saved session

• Going to use a proxy for this account? Pass it now, so the login and later uploads share one IP: autotok login -n my_account -p http://user:pass@host:port (see Proxies).
• Repeat with a different -n name for each account.
• Sessions are saved in ~/.autotok/accounts. Treat them like passwords.

autotok upload -u my_account -v "C:\Videos\clip.mp4" -t "My caption #fyp"

That's it. More examples:

autotok upload -u my_account -v clip.mp4 -t "Just for me" -vi 1                 # private
autotok upload -u my_account -v clip.mp4 -t "Later #fyp" -sc 3600               # publish in 1 hour
autotok upload -u my_account -yt "https://www.youtube.com/shorts/xxxxxxxxxxx" -t "Caption"

-v takes a path to any video, or just a file name if the video is in ~/.autotok/videos. See the CLI reference for every option.

Settings such as a cloud browser or its API key go in a .env file. autotok looks in the folder you run it from, then in ~/.autotok/.env; the second location works from any folder. Copy .env.example (from this repository) or start with just what you need:

# ~/.autotok/.env   (Windows: C:\Users\<you>\.autotok\.env)
AUTOTOK_BROWSER=browserbase
BROWSERBASE_API_KEY=bb_live_...

• One-command uploads through TikTok's web upload API. No clicking through the TikTok website.
• Proxy per account: login, request signing and every upload go through the same proxy, so each account always appears from one IP. HTTP(S) and SOCKS5 are supported.
• Multiple accounts: log in once per account and upload by name.
• Scheduling: TikTok-side scheduling from 15 minutes to 30 days ahead (TikTok limits some accounts to 10 days). The self-hosted web app adds a scheduler with no time limit.
• Post settings: public or private, and comments, duets, stitches and the AI-generated label on or off.
• Clickable hashtags and @mentions, emoji-safe.
• YouTube links download automatically with yt-dlp.
• Python SDK, CLI, and an optional web app (Docker) with a browser-based login.
• MCP server so AI agents (Claude, Cursor, VS Code...) can upload and schedule for you.
• Cloud browsers: run the login and signing browser on Browserbase or Steel (cloud or self-hosted) instead of your computer.
• Docker image for running the CLI on servers.
• No Node.js: TikTok's request signatures are computed in headless Chromium.

Running several accounts from one IP is the quickest way to get them flagged. autotok lets you give every account its own proxy, and uses it everywhere that account talks to TikTok:

Set, change and test proxies

autotok setup asks for the proxy, tests it and logs in through it. To do it by hand:

autotok login -n alice -p http://user:pass@gate.example.com:7000   # set at login
autotok proxy set alice socks5://user:pass@gate.example.com:1080     # set or change later
autotok proxy test -u alice                                          # shows the IP TikTok will see
autotok proxy clear alice                                            # go direct
autotok accounts list                                                # every account and its proxy (passwords hidden)

autotok upload -u alice -v clip.mp4 -t "hi" -p http://other:pass@host:8000   # one-off override
autotok upload -u alice -v clip.mp4 -t "hi" --no-proxy                       # ignore the saved proxy once

Accepted formats (scheme defaults to http)

http://user:pass@host:port      https://user:pass@host:port
socks5://user:pass@host:port    host:port
user:pass@host:port             host:port:user:pass      ← the export format most providers use

Tips

• Use residential or mobile proxies, one per account, with a sticky session (same IP for hours). A proxy that rotates the IP on every request makes a logged-in session look hijacked.
• Log in through the proxy you will upload with (autotok login -p ...). Logging in from home and uploading from a proxy is a mismatch TikTok can see.
• Chromium can't use SOCKS5 proxies that need a username and password. For the login step, use your provider's HTTP endpoint (uploads work with either).
• Passwords are never printed: the CLI and web app show http://user:****@host:port.

Need proxies? See our sponsors.

autotok needs a browser in two places: the one-time login, and signing each upload (TikTok's anti-bot signatures are computed in a browser page). By default that's Chromium on your machine. Set AUTOTOK_BROWSER to run it in the cloud instead. Nothing then needs to be installed or opened locally, which makes it a good fit for servers and Docker.

Put the settings in your .env file (copy .env.example):

AUTOTOK_BROWSER=browserbase
BROWSERBASE_API_KEY=bb_live_...

Then use autotok as usual:

autotok browser check                 # starts a session and shows the IP it browses from
autotok login -n alice -p http://user:pass@gate.example.com:7000
autotok upload -u alice -v clip.mp4 -t "Hello #fyp"

Or pick per command: autotok --browser steel login -n alice.

Logging in: autotok login prints a live-view link and opens it in your normal browser. Log in to TikTok in that page; the session is saved as soon as you're in and the cloud browser is closed.

Self-hosted Steel is free and open source:

docker run -p 3000:3000 -e DOMAIN=localhost:3000 ghcr.io/steel-dev/steel-browser

and in .env:

AUTOTOK_BROWSER=steel
STEEL_BASE_URL=http://localhost:3000

Set DOMAIN to the address you reach the server at. Without it, steel-browser tells browsers to connect to 0.0.0.0 and its live view stays blank, so you can't log in. autotok warns when it sees this. A self-hosted steel-browser runs one session at a time.

Good to know:

• Use an account proxy with cloud logins. The account's proxy is applied to the cloud login session, so TikTok sees the same IP at login and upload. Without one, TikTok sees the login come from the provider's IP and uploads come from yours.
• "Verify it's really you" after logging in means TikTok saw an unfamiliar device on a datacenter IP. You can complete it in the live view, but it's better to avoid it: use an account proxy, or on a paid Browserbase plan set BROWSERBASE_PROXY_COUNTRY=GB (your country) to log in through a home IP there. On the free plan, BROWSERBASE_REGION set to the region nearest you (e.g. eu-central-1) at least avoids a US login.
• The live view shows only the TikTok tab, so autotok closes popups such as "Continue with Google" during cloud logins. Use the QR code, email/username or phone instead.
• Browserbase only takes HTTP/HTTPS proxies, and custom proxies need its Developer plan or higher. Custom proxies on Steel Cloud may need a paid plan too.
• Only the browser runs in the cloud. The video file still goes from your machine (through the account proxy) straight to TikTok, so cloud usage stays small: one short session per login and per upload.
• Want another provider? Subclass autotok.browsers.CloudProvider and call autotok.browsers.register_provider("name", YourProvider).

import autotok

autotok.login("alice", proxy="http://user:pass@gate.example.com:7000")   # once; opens a browser

result = autotok.upload_video("alice", "clip.mp4", "Hello #fyp")
print(result.video_id)

# More control: reuse a client, set post options, schedule
from datetime import datetime, timedelta, timezone

client = autotok.Client.from_account("alice")          # uses alice's saved proxy
client.upload(
    "clip.mp4",
    "Going live tomorrow #fyp @friend",
    schedule=datetime.now(timezone.utc) + timedelta(days=1),   # or seconds, or a timedelta
    visibility="public",            # or "private"
    allow_comment=True,
    allow_duet=False,
    allow_stitch=False,
    ai_label=False,
)

# Errors are typed; `retryable` is True only when nothing reached TikTok's publish step.
try:
    client.upload("clip.mp4", "caption")
except autotok.PublishError as e:          # TikTok rejected the post (e.g. guidelines)
    print(e.status_code, e.status_msg)
except autotok.AutotokError as e:
    print("failed:", e, "safe to retry:", e.retryable)

Already have a sessionid cookie from your browser? Skip the login window:

autotok.import_session("alice", "<sessionid value>", datacenter="<tt-target-idc value>")

autotok includes an MCP server, so AI agents such as Claude, Cursor or VS Code can upload and schedule videos for you. It uses the same saved accounts as the CLI.

uv tool install "autotok[youtube,mcp]"    # or: pip install "autotok[youtube,mcp]"

Claude Code

claude mcp add autotok -- autotok mcp

Claude Desktop, Cursor and other clients: add this to the client's MCP settings (claude_desktop_config.json, .cursor/mcp.json, ...):

{
  "mcpServers": {
    "autotok": { "command": "autotok", "args": ["mcp"] }
  }
}

If the client can't find autotok, use the full path that which autotok (Windows: where autotok) prints. Then ask, for example: "Upload clip.mp4 to my_account with the caption 'Hello #fyp', scheduled for 6 pm tomorrow."

• Your sessions stay on your computer. The tools never return cookies, and proxy passwords are masked.
• Every upload is a real post. Most clients ask before each tool call; keep that on for upload_video. An upload error that says the post may or may not exist tells the agent not to retry, so nothing is posted twice.
• Streamable HTTP: autotok mcp --http serves http://127.0.0.1:8000/mcp (--host, --port to change). There is no authentication, so keep it on localhost or behind your own.

autotok setup                         # step by step: browser, proxy (tested), TikTok login
autotok login     -n NAME [-p PROXY] [--force] [--sessionid ID --datacenter DC]
autotok upload    -u NAME (-v FILE | -yt URL) -t CAPTION [options]
autotok accounts  list | check NAME | remove NAME
autotok proxy     set NAME PROXY | clear NAME | test [PROXY | -u NAME]
autotok show      -u (accounts) | -v (videos)
autotok browser check [-u NAME]       # test the local or cloud browser
autotok install-browser [--with-deps]
autotok shell                         # interactive prompt
autotok mcp [--http [--host H] [--port P]]   # MCP server for AI agents (needs autotok[mcp])

autotok --browser {local,browserbase,steel} <command> ...   # choose where the browser runs

Upload options:

python cli.py ... in this repository still works and is the same as autotok ....

The Dockerfile in the repository builds the CLI with Chromium included:

docker build -t autotok .

docker run --rm -it -v autotok-data:/data autotok accounts list
docker run --rm -it -v autotok-data:/data -v "$PWD/videos:/data/videos" \
  autotok upload -u alice -v clip.mp4 -t "Hello #fyp"

Saved accounts live in /data, so keep it in a volume. Containers have no screen, so log in with a cloud browser and open the link it prints:

docker run --rm -it -v autotok-data:/data \
  -e AUTOTOK_BROWSER=browserbase -e BROWSERBASE_API_KEY=bb_live_... \
  autotok login -n alice -p http://user:pass@gate.example.com:7000

You can also save a session cookie from your own browser with login --sessionid ... --datacenter ....

Using a cloud browser for everything? Build without the local Chromium for a much smaller image: docker build --build-arg INSTALL_BROWSER=false -t autotok:slim .

A local control panel with accounts, proxies, a video library, uploads and a scheduler, plus a virtual browser for logging in. Requires Docker.

git clone https://github.com/makiisthenes/TiktokAutoUploader.git
cd TiktokAutoUploader
docker compose up --build

Open http://localhost:3000. The API docs are at http://localhost:8000/docs.

• Accounts: add accounts by logging in through the built-in virtual browser (with an optional proxy), then set, change or test each account's proxy and check its session.
• Upload: drop a file, pick one from the library, or paste a YouTube link. Publish now or schedule for any time.
• Schedules: the scheduler retries only failures that happened before anything reached TikTok, so a job can't post twice.

Data lives in ./data. To use an account you logged in to with the CLI, copy its file into the app's store and click Import from disk on the Accounts page:

mkdir -p data/autotok/accounts && cp ~/.autotok/accounts/my_account.json data/autotok/accounts/

(Files the containers write are owned by root on Linux, so use the web app's own login for accounts you manage there.)

To sign uploads with a cloud browser, put AUTOTOK_BROWSER and its settings (see Cloud browsers) in a .env file next to docker-compose.yml. Logging in through the web app still uses its built-in virtual browser.

Scripts that call the API directly must send the header X-Requested-With: autotok on POST/PATCH/DELETE requests. This stops other web pages from using the local API.

The web app has no login of its own and only listens on 127.0.0.1. Don't expose it to the internet.

Everything is optional. Put settings in a .env file: copy .env.example to .env and fill in what you use. autotok reads .env from the folder you run it in, then ~/.autotok/.env. Variables already set in your environment take priority, and docker compose reads the same file. .env holds API keys, so never commit it (it's in .gitignore).

Account files contain your TikTok session (equivalent to your password). They are saved with owner-only permissions; keep them private.

• Install: follow Getting started (uv tool install "autotok[youtube]", then autotok install-browser). Node.js and npm install are no longer needed.
• Sessions: saved sessions in CookiesDir/ are migrated automatically the first time you use them from the repository folder. The flags are the same, and python cli.py still works.
• Scheduling: -sc schedules on TikTok's side again (15 minutes to 10 days). The 1.x local queue (cli.py schedule list|cancel) now lives in the web app's Schedules page.
• Bug fixes: privacy, comment, duet and stitch settings are now actually applied. Earlier versions always posted publicly with everything allowed. Hashtags are clickable again.
• Removed: config.txt and the unused ImageMagick video-editing helpers.

See also Common problems.

Video showcases main usage of the app, uploading a video to TikTok.

Fill waiting list form: https://forms.gle/M4KpdfruqCukQvj99

If you are looking for something more, which can get you faster to your goal, I offer software which can:

• Handle more than 1000 accounts
• Upload identical videos to multiple accounts automatically
• Schedule videos for multiple accounts, 20 days to 2 years in advance
• Automatically source videos from YouTube, X, Reddit, TikTok
• Setup uploading pipelines, from source to uploading schedule
• Metrics for viewing current performance of these different accounts
• Personalised support from me for any issues you may face for up to 3 months
• Proxy support, clean and modern UI

Available for purchase, if interested please email me at michaelperes562@gmail.com with subject line Tiktok Bot Software or else I might miss the email.

autotok is dual-licensed:

• Open source: GNU AGPL v3.0. Use, modify and share it freely. If you distribute a modified version, or run one as a service for others, you must publish your source code under the same licence.
• Commercial: to build autotok into closed-source products or paid services without the AGPL obligations, or to use autotok Pro, contact michaelperes562@gmail.com.

Copyright © 2021–2026 Michael Peres. Releases before 2.0.0 remain under the licence they were published with. The bundled signing scripts are third-party code, see THIRD_PARTY_NOTICES.md. Contributions are welcome under the terms in CONTRIBUTING.md.

If you like the work provided, please consider supporting me through Patreon or Ko-Fi, or star the project.

• Thanks @DelvinBa for updating to TikTok's new upload endpoint. (09/12/2024)

Please report bugs in the issues tab rather than by email.

This project is not affiliated with TikTok. Automating uploads may break TikTok's terms of service and can get accounts restricted or banned. Only upload content you have the rights to. Use at your own risk.