Skip to main content

Develop Cloud Functions with AI IDE

With an AI IDE, natural-language conversation drives the full cloud-function workflow: the AI generates code, installs dependencies, deploys, and inspects logs — removing the need to switch between console and command line.

This guide shows how to integrate the CloudBase AI Toolkit and drive creation, update, deployment, log query, and HTTP access configuration of cloud functions through conversation inside an AI IDE.

tip

WorkBuddy and CodeBuddy ship with a built-in Tencent CloudBase connector — one click to enable, no manual MCP configuration required. Other IDEs need to wire up the CloudBase MCP their own way.

Use cases​

  • From zero: describe the requirement; the AI generates the code skeleton and deploys it to your environment
  • Iterate: update business logic, environment variables, timeout / memory settings, then redeploy
  • Troubleshoot: list functions, view details, tail recent execution logs
  • Expose publicly: attach an HTTP access path for your function

For selection conventions and more prompt patterns, see the Cloud Functions Skill.

Prerequisites​

1. Connect to CloudBase​

Pick the IDE you use and follow its setup guide:

IDEIntegrationSetup Guide
WorkBuddyBuilt-in CloudBase connector, one-click enableWorkBuddy Setup Guide
CodeBuddyBuilt-in CloudBase integrationCodeBuddy Setup Guide
CursorManual MCP configurationCursor Setup Guide
Claude CodeManual MCP configurationClaude Code Setup Guide
Other IDEsGeneric MCP wiringFull IDE List

WorkBuddy: one-click connector​

WorkBuddy ships with the Tencent CloudBase MCP connector out of the box:

  1. In the chat panel, click the Connectors button at the bottom-left of the input box
  2. Locate Tencent CloudBase, click Connect to enable it
  3. In the chat input, type /cloudbase to pick the CloudBase Skill and describe what you want

For a walkthrough with screenshots, see the WorkBuddy Setup Guide.

Install the Cloud Functions Skill (recommended)​

The Cloud Functions Skill teaches the AI to generate code that follows platform conventions, correctly distinguishing Event Functions from HTTP Functions and handling scf_bootstrap, port 9000, and related runtime requirements.

Install all CloudBase Skills:

npx skills add tencentcloudbase/cloudbase-skills

Or install only the Cloud Functions Skill:

npx skills add https://github.com/tencentcloudbase/skills --skill cloud-functions

View the Skill online: cloud-functions.

2. Start chatting​

Once signed in and bound to an environment, describe your requirement directly, for example:

  • "Create a cloud function called hello_world and deploy it to the current environment."
  • "Update hello_world to read name from the request body and return hello, {name}. Redeploy."
  • "Show me the last 10 execution logs of hello_world."
  • "Attach an HTTP access path /hello to hello_world."
  • "Bump hello_world timeout to 30s and memory to 512MB."

FAQ​

MCP fails to connect, or the tool count shows 0​

Check the config file format, restart the IDE, and verify network connectivity. See the AI Toolkit FAQ for details.

Function created but calls fail with EXCEED_AUTHORITY​

New environments disable anonymous login by default, and default function security rules reject anonymous callers. If your function needs public access, ask the AI to update the function security rule to allow the intended caller scope — do not rely on anonymous login.

Should I still use in-console editing​

Yes. The console suits two-line tweaks and quick validation; the AI IDE suits long sessions and full-feature development. Treat the deployed cloud version as the source of truth — when both paths edit the same function, pull the cloud version back as the baseline.