Skills
What is Skills
Skills is a community-open specification for injecting domain-specific professional expertise and operational processes into AI Agents. Skills exist as structured Markdown files, containing complete task flows, judgment logic, and exception handling mechanisms. After loading Skills, an AI Agent can autonomously complete professional tasks by following predefined processes.
Currently, ALL AI programming tools that support the Skills Mechanism can use Skills, including but not limited to Claude Code, CodeBuddy, and Cursor.
Included Skills
edgeone-makers-tools organizes the following eight Skills by domain. The AI Agent automatically matches and loads them based on the task.Skill | Coverage |
makers-agents | AI Agent development (DeepAgents, LangGraph, Claude Agent SDK, OpenAI Agents, CrewAI) |
makers-edge-functions | Edge Functions (V8 lightweight runtime) |
makers-cloud-functions | Cloud Functions(Node.js / Go / Python) |
makers-storage | KV and Blob storage |
makers-middleware | Middleware (authentication, rewrite, routing, and so on) |
makers-deploy | Deploy projects to EdgeOne. |
makers-cli | EdgeOne CLI command reference |
makers-recipes | Project structure templates and scaffolding |
Installation Method
npx skills add TencentEdgeOne/edgeone-makers-tools
You can also install via natural language. In an AI conversation, provide the repository address to the Agent, for example, "Help me install this skill:
https://github.com/TencentEdgeOne/edgeone-makers-tools". The Agent will automatically pull the repository and complete the installation. Manual installation is also supported (download the Skill file and place it in the corresponding directory of the tool).Usage Methods
After installation, simply describe your requirements in natural language. The AI Agent will automatically determine which capability to use and execute it. It is recommended to include specific capability / platform / framework keywords in your description (such as "EdgeOne Makers", "Edge Functions", "Cloud Function", "Agent", or "middleware") for more precise triggering. Overly broad descriptions may fail to accurately match the corresponding Skill.
Example:
Scenario | Example prompt |
Deployment | Deploy this Next.js project to EdgeOne Makers and provide me with the preview link. |
AI Agent | Write a conversational Agent on EdgeOne Makers using the Claude Agent SDK. |
Edge Functions | Implement a page visit counter using EdgeOne Edge Functions + KV. |
Cloud Functions | Write an EdgeOne Python Cloud Function using FastAPI. |
Middleware | Write an EdgeOne Makers middleware to add authentication to the /api route. |
Scaffolding | Use edgeone makers create --template openai-agents-starter-node to pull and run an Agent template. |
After being triggered, the AI Agent automatically completes the corresponding process:
Development: Analyze requirements (determine whether to use an Agent framework, Edge/Cloud Functions, or middleware) → Generate code according to platform specifications → Guide local debugging using
edgeone makers dev (default port 8088).Deployment: Environment check (CLI installation/version) → Login authentication (China/Global) → Project build → Deployment and upload → Return the online URL and console address.
During the process, the AI Agent may ask you to confirm information (such as selecting the China site/international site or providing an API Token). Simply respond by following the prompts.
Supported Project Type
Supports all EdgeOne Makers project types, including:
AI Agent projects: Support Agent applications built on mainstream Agent frameworks (such as Claude Agent SDK, OpenAI Agents, LangGraph, and so on), which can be delivered within the same project as Web.
High-performance full-stack projects: Support SSR and ISR features of frameworks like Next.js and Nuxt, integrate edge and cloud computing capabilities, and easily build complex integrated frontend-backend projects.
Static and dynamic websites: Supports websites built with static site generators (such as Astro, Hexo, and so on) and single-page applications built with modern frontend frameworks (such as React, Vue, and so on).
The CLI automatically identifies the project framework and selects the correct build and deployment policy, eliminating the need for manual configuration.
Login Method
Skills support two login methods. The AI Agent will automatically select based on your runtime environment.
Environment | Login Method | Description |
local desktop environment | browser login | Open browser and complete login automatically |
remote server | API Token Login | Guide you to create a Token through the console and enter |
Note:
How to choose China site/international site: It depends on which site your EdgeOne account belongs to. If you use a Tencent Cloud China site (
console.cloud.tencent.com) account, select "China". If you use an international site (edgeone.ai) account, select "Global". The account systems for the two sites are independent. Simply choose based on the site you actually signed up for and use.API Token get address:
Note:
API Token has account level permission, keep it safe and never submit it to the code repository.
