Guide · Agent skill
How to use the TikTok Video Publisher agent skill
Updated · written with AI from the project's own documentation and checked against our catalogue data
The tiktok-upload skill is a set of instructions (a SKILL.md file) that teaches an AI agent how to publish a single video to TikTok using the social-auto-upload repository's Chrome-based uploader. It is written for people who already run that repository locally and want an agent to handle the posting steps. After following this guide you will know what to install, what your machine needs, how to ask for a post in plain language, and what the agent will check before it publishes.
Quick start
- 1Set up the repository. Clone social-auto-upload locally and install its Python dependencies. Configure Chrome with Patchright or Playwright in conf.py.
- 2Log in to TikTok yourself. Use examples/get_tk_cookie.py as a guide to log in on the TikTok page. The session is stored in cookies/tk_uploader/account.json.
- 3Download the skill ZIP. Get the tiktok-upload ZIP from the page where you found this guide. It contains SKILL.md.
- 4Install the skill. Unzip it into ~/.claude/skills/ for personal use or into .claude/skills/ in a project. Alternatively, upload it in the Claude apps under Settings → Capabilities → Skills.
- 5Prepare your video. Place a local MP4 file where you can reference its path, such as videos/demo.mp4 in the repository.
- 6Ask for one post. Describe the video, title, tags, and timing in plain language, then confirm each detail when the agent asks.
What the skill makes the agent do
Without the skill, an agent has no reliable knowledge of how to post to TikTok from this repository, and it may guess at commands. With the skill installed, the agent follows the repository's own uploader, uses a login session stored locally, and asks you to confirm the details before anything goes live.
The skill does not call a TikTok API. It automates browser actions through the repository's Chrome uploader, so it depends on your logged-in browser session and on TikTok's web page layout.
- Publishes one video at a time with a title, tags, and either immediate or scheduled posting
- Asks you to confirm the account, video, title, tags, and posting time before it runs
- Reports success only after TikTok's page confirms the post
What you need before you install
The skill assumes a working setup. It does not install the repository for you, so check each item first.
- A local clone of the social-auto-upload repository, with its Python dependencies installed
- Patchright or Playwright with a Chrome installation, configured in conf.py
- A TikTok account that you log in to through your own browser session, stored in cookies/tk_uploader/account.json
- A local MP4 video file
Install the tiktok-upload skill
Download the skill ZIP from the page where you found this guide, then choose one of two install locations. The skill package contains a single file, SKILL.md.
For a personal install, unzip the package into ~/.claude/skills/. For a project-only install, unzip it into the .claude/skills/ folder inside that project. You can also upload the ZIP in the Claude apps under Settings → Capabilities → Skills.
- Personal: ~/.claude/skills/
- Project only: .claude/skills/ inside your project folder
- Claude apps: Settings → Capabilities → Skills, then upload the ZIP
Log in once, without sharing your cookies
The account session is kept in cookies/tk_uploader/account.json. For the first login, the repository includes examples/get_tk_cookie.py, which the skill points to. You complete the login yourself on the TikTok page.
Do not ask the agent to read, copy, or commit the cookie file. Keep it out of any shared repository or backup.
Ask the agent to publish a video
Once the skill is installed, you can describe the post in ordinary language. The agent should load the skill and then ask you to confirm the account, the single video, the title, the tags, and whether to post now or schedule it.
Example requests you might type:
- Publish videos/demo.mp4 to TikTok now with the title I give you and the tags I list, and confirm everything before posting.
- Schedule one TikTok video for a specific date and time I will give you. Ask me to confirm before you run it.
- Check that my TikTok login is still active before you upload anything.
The repository code the agent uses for one video
For a single video, the skill follows the repository API flow shown below. The value 0 for publish_date means publish immediately. Scheduled posts need an explicit datetime that you provide; the agent should not work out a time on its own. The TiktokVideo class has no separate description parameter.
import asyncio
from pathlib import Path
from conf import BASE_DIR
from uploader.tk_uploader.main_chrome import TiktokVideo, tiktok_setup
account_file = Path(BASE_DIR) / "cookies" / "tk_uploader" / "account.json"
video_file = Path(BASE_DIR) / "videos" / "demo.mp4"
asyncio.run(tiktok_setup(str(account_file), handle=True))
app = TiktokVideo("标题", str(video_file), ["标签"], 0, str(account_file))
asyncio.run(app.main())Avoid the bundled batch script unless you ask for it
The repository also contains examples/upload_video_to_tiktok.py. It processes every MP4 in a folder, extracts titles and tags from the files automatically, and schedules posts for 16:00 the next day. Those behaviours skip per-item review.
The skill advises against running this script unless you explicitly request that behaviour. If you want more than one video, ask the agent to go through them one at a time with your confirmation for each.
What the agent checks and how it handles problems
Before each publish, the agent should confirm the details with you. If the TikTok page layout or login state does not match what the skill expects, the agent should stop and report the problem rather than push ahead.
If a publish ends with an unknown result, the agent should check TikTok Studio first. It should not post the video again straight away, because that could create a duplicate.
- Stop and report if the page layout or login state is not as expected
- Report success only after the platform page confirms the submission
- On an unknown result, check TikTok Studio before any retry
What people use it for
- Posting one finished video to your own TikTok account with a title and tags you approve
- Scheduling a single video for a specific time you choose, with a confirmation step before it is queued
- Checking that the saved login is still working before a planned post
- Keeping a record of what was published by having the agent report only confirmed results
Frequently asked questions
Is the tiktok-upload skill free to use?
The listing states the licence is MIT. The SKILL.md front matter, however, names MIT-0, so check the licence in the repository before you reuse or redistribute the skill. The source does not describe any cost.
What do I need before I can use this skill?
You need a local clone of the social-auto-upload repository with its Python dependencies, Patchright or Playwright with a Chrome installation configured in conf.py, a TikTok account logged in through your own browser session, and a local MP4 file.
Does this skill use the TikTok API?
No. The skill is a guide for automating browser-based uploads with the repository's Chrome uploader. It is not a TikTok API client, and the source notes that TikTok has no sau tiktok command.
Where do I install the tiktok-upload skill?
Unzip the downloaded ZIP into ~/.claude/skills/ for personal use, or into .claude/skills/ inside a project. You can also upload it in the Claude apps under Settings → Capabilities → Skills.
Can the agent post many videos at once?
The skill advises against it by default. The bundled batch script auto-extracts titles and tags and schedules posts without per-item review, so it should only run when you explicitly ask for that behaviour.
What happens if the upload fails or the result is unclear?
The agent should stop if the page layout or login state is unexpected. For an unknown result, it should check TikTok Studio first and must not repost immediately, to avoid duplicates.
New agent skills, repos and AI tools — every week
The best new skills and open-source projects, the fastest-growing repos, and one free prompt. Five minutes to read. Unsubscribe anytime.
Our verdict
This skill suits creators and marketers who already run social-auto-upload locally, own the TikTok accounts they post to, and want an agent to post single videos with a confirmation step. It is a poor fit for anyone who does not control the account, or for bulk or unattended posting, since it automates browser actions that may breach TikTok's terms and risk restrictions on the account.
Download the skill