Andromity loads skills from folders you trust. Install with andromity skill install cloudflare-agents-sdk — autonomy and approval limits stay configurable per session.
Destination: Paste into any AI Agent (Cursor, Claude, Windsurf, AGY, Andromity)prompt mode
# Install the "Cloudflare Agents SDK" skill
You are installing an AgenticMarket skill into this workspace. Follow these steps exactly.
## 1. Detect your environment and choose the destination file
- Andromity: `.andromity/skills/cloudflare-agents-sdk/SKILL.md` (or user scope `~/.andromity/skills/cloudflare-agents-sdk/SKILL.md`)
- Antigravity / Gemini CLI: `.agent/skills/cloudflare-agents-sdk/SKILL.md`
- Cursor: `.cursor/rules/cloudflare-agents-sdk.mdc`
- Claude Code: append to `CLAUDE.md` (or `.claude/skills/cloudflare-agents-sdk.md`)
- Windsurf: append to `.windsurfrules`
- Cline / Roo Code: `.clinerules/cloudflare-agents-sdk.md`
- GitHub Copilot: append to `.github/copilot-instructions.md`
- ChatGPT / web only: paste the content into Custom Instructions
- Any other agent: save to `.agent/skills/cloudflare-agents-sdk/SKILL.md`
## 2. Create the destination file
- Preserve every existing file and configuration entry.
- Create parent folders as needed.
- Never overwrite unrelated content.
## 3. Write exactly this content
```markdown
---
name: cloudflare-agents-sdk
description: "Build stateful AI agents, durable workflows, real-time WebSocket communication, and MCP servers on Workers."
author: cloudflare
license: mit
version: 1.0.0
---
# Cloudflare Agents SDK
Build robust, stateful AI agents utilizing Cloudflare Workers, Durable Objects, and Vectorize.
## Features
- Stateful conversation and persistent memory across disconnects
- Durable workflows that survive server restarts and network interruptions
- Callable RPC methods and type-safe client React hooks
- Native Model Context Protocol (MCP) server endpoints
## Example Agent Class
```typescript
import { Agent } from "@cloudflare/agents";
export class ChatAgent extends Agent<Env> {
async onConnect(connection: Connection) {
connection.send(JSON.stringify({ type: "ready", agentId: this.id }));
}
async onRequest(request: Request): Promise<Response> {
return new Response("Agent online: " + this.id);
}
}
```
```
## 4. Verify
- Confirm the file exists and is readable.
- Report the destination path and confirm the "Cloudflare Agents SDK" skill is installed and active.
Skill metadata:
- Slug: cloudflare-agents-sdk
- Category: ai-agents
- License: mit
- Source: AgenticMarket