THE WORKFLOW
Your agent does the reasoning.
Your browser does the browsing.
The skill brings X Home and Search collection, replies, reposts, quote-posts, original posts, and thread interactions into one agent workflow. Your host agent selects and drafts; browser scripts carry out the reviewed actions over CDP.
Remote Browser supplies the authenticated browser session. The skill connects to that session rather than launching a separate local browser.
GET STARTED
From CDP to your first dry run.
- 01
Prepare your browser
Create a Remote Browser session and sign in to the intended X account. Keep that session available and copy its CDP endpoint.
- 02
Install the skill
Follow the repository installer and choose Codex, Claude, Hermes, or OpenClaw. Node.js 20 or newer and npm are required.
- 03
Connect and set your focus
Invoke x-community-automation in your agent. In the first-run conversation, provide your CDP endpoint, topics, allowed actions, language, and limits. The skill stores the configuration locally.
- 04
Review, approve, verify
Start with a dry run to review the account, candidate posts, and proposed actions. Approve the actions you want to publish, then check each result against its source post.
git clone https://github.com/HuberTRoy/x-automation.git
cd x-automation
# Review the installer, then choose your agent platform.
bash install.sh --platform codexChoose a different platform in the installer if you use Claude, Hermes, or OpenClaw. See the installation guide for all options.
Use x-community-automation with my Remote Browser CDP session. Collect X posts relevant to my configured focus and prepare a dry run with candidate posts and suggested replies. Show me the results for review before publishing anything.
Keep the context between runs.
Use a persistent browser profile for your signed-in account. The skill can reuse its configured focus and action limits, verify published results under their source posts, and optionally report progress and completion to Telegram.
Read about browser profiles