AI Dialogue Deployment: Deploy Project with One Sentence Using Skill
This article introduces how to install EdgeOne Pages Deploy Skill to complete project deployment with a single command in any AI programming tool, simplifying the complex deployment process and achieving "dialogue as deployment".
Overview
Skill is an open standard proposed by Anthropic, using structured Markdown files as the carrier to inject domain-specific professional knowledge and operation procedures into AI tools. When an AI tool loads a Skill, it gains the ability to autonomously complete professional tasks according to predefined processes. You can think of a Skill as a "user manual" for AI—the Skill itself does not perform any operation, but tells the AI tool what to do, how to do it, and how to handle problems when encountered.
EdgeOne Pages Deploy Skill is exactly such a professional user manual written for deployment scenarios. It encapsulates the complete deployment process, environment judgment logic and error handling mechanism. Any AI tool supporting Skill (such as CodeBuddy, Claude Code, Cursor) can autonomously determine environment, select policy and execute deployment like an operation and maintenance engineer after loading it. Moreover, projects deployed to EdgeOne Pages are persistent—every deployment updates the same site URL, allowing continuous iteration just like using traditional CI/CD process. Projects can also be managed in the EdgeOne Pages console to bind custom domain, view deployment history and more.
Installing a Skill
Skill installation is super simple. Below are three methods.
Method One: Install via Natural Language (Recommended)
Enter "Help me install this skill: https://github.com/edgeone-pages/edgeone-pages-skills" in any AI tool supporting Skill.
The AI tool will automatically pull the repository, parse the Skill file and complete installation with no need for manual intervention.
Method Two: Install via Command Line
Execute in the terminal:
npx skills add edgeone-pages/edgeone-pages-skills
Method Three: Manual Installation
Download the Skill file from the GitHub repository, and place the
skills/edgeone-pages-deploy/ directory into the corresponding rule directory for the AI tool. For the directory location and detailed setting method of each tool, refer to the following text "for use in different AI tools".
Deploying a Project
After installing the Skill, you can trigger deployment via natural language in AI conversation. The following demonstrates two typical scenarios.
Scenario One: Develop and Deploy Project in CodeBuddy
When you need to add a feature in your EdgeOne Pages project and want to develop and deploy it to the preview environment as soon as possible for verification.
Step 1: Initiate a deployment request in an AI dialogue
Input in the AI dialog box of CodeBuddy: "Help me create a dark mode, then deploy a preview environment to EdgeOne Pages."
Step 2: Automatically perform environment check, logon and deployment
After loading the Skill, CodeBuddy will automatically follow the steps below to execute:
1. Environment check: Detect whether EdgeOne CLI is installed with the correct version. If not installed, auto-execute
npm install -g edgeone.2. Login authentication: Detect the login status. If not logged in, guide you through authentication via a browser.
3. Project build: Automatically identify the project framework (such as Vue.js) and execute the build command.
4. Deploy upload: Deploy build artifacts to EdgeOne Pages.
During the entire process, you just need to confirm a small amount of information when seeing the prompt (if selected China site or international site), and the rest will be automatically performed.
Step 3: Get Deployment Result
After successful deployment, AI will be returned:
Access URL: You can directly open it in the browser.
Console URL: Used for project management in the EdgeOne Pages console.
From initiating deployment to obtaining the online URL, it usually takes just 1-2 minutes. Afterward, each time you modify code, just say "Deploy to EdgeOne Pages" again to push the update online.
Scenario Two: Create and Deploy Project from Scratch in OpenClaw
OpenClaw is an open source AI Agent framework that supports access to enterprise WeChat, Lark, Telegram and other communication tools, allowing you to have AI conversations directly in mobile phone IM to complete development and deployment.
Suppose you get inspired on the subway and want to make a mini game to share with friends. You don't need to turn on the computer, just say one sentence in IM: "Make a Snake Game for me and deploy it to EdgeOne Pages."
The AI will generate game code from scratch, then automatically trigger the deployment process: check environment → login authentication → build project → deploy online, ultimately returning an accessible online URL. Share the link with friends, and they can play it right away. Even if the game encounters a bug, you can directly ask OpenClaw to fix it and redeploy inside, then you will see the fixed version.
In a remote server, SSH, or CI/CD environment, unable to complete login through a browser, Skill will automatically switch to API Token authentication. Token can be created at the following address:
The Agent will also ask whether to save the Token to local (
.edgeone/.Token) after first use, and subsequent deployments can automatically reuse it without repeat input.Note:
API Token has account level permission for Pages. Keep it safe and do not submit it to the code repository.
Using in Different AI Tools
Pages Deploy Skill is based on common Markdown standard and fully compatible with ALL AI programming tools that support Anthropic Skills standard.
Claude Code
Claude Code has native support for Skill standard. After installation via command line or natural language, Skill will be automatically placed in the
~/.claude/skills/ (global) or .claude/skills/ (project level) directory. You can also manually copy the Skill directory to these locations.CodeBuddy
CodeBuddy IDE has native support for Skill installation. After installation via command line or natural language, Skill will be automatically placed in the
.codebuddy/skills/ directory. You can also manually manage Skill through Settings → Skills. For more information, see Using CodeBuddy IDE.Cursor
Cursor has native support for Skill installation. After installation via command line or natural language, Skill will be automatically placed in the
~/.cursor/skills/ directory. You can also manually copy the Skill directory to this location.
FAQs
Q: What to do if the deployment fails?
A: Deploy Skill has built-in error handling logic. After loading the AI tool, it will automatically attempt to fix errors based on the error type. Common automatic fix scenarios include: CLI version mismatch (auto reinstall the correct version), login expired (guide to re-login), build failed (analyze build log and provide fix advice). If the issue remains unresolved, you can submit the error information to the AI for further analysis.
Q: Which project frameworks are supported?
A: Support all EdgeOne Pages compatible frameworks, including Next.js, Nuxt, Astro, Vite, React, Vue, SvelteKit. EdgeOne CLI will automatically identify the framework type and select the correct build policy without manual configuration.
