Anthropic Claude API ও Prompt Caching — Prompt Engineering & AI Tools

🔑 Claude API Setup

npm install @anthropic-ai/sdk
ANTHROPIC_API_KEY=sk-ant-...

💬 Basic Call

import Anthropic from "@anthropic-ai/sdk";
const client = new Anthropic();

const msg = await client.messages.create({
  model: "claude-sonnet-4-20250514",
  max_tokens: 1024,
  system: "You are a helpful assistant.",
  messages: [
    { role: "user", content: "Hello" }
  ]
});
console.log(msg.content[0].text);

⭐ Prompt Caching (90% off!)

Long system prompt বা document — repeat call-এ massive cost savings।

messages: [{
  role: "user",
  content: [
    {
      type: "text",
      text: "[Huge document, 10K tokens]",
      cache_control: { type: "ephemeral" }
    },
    {
      type: "text",
      text: "Question: ..."
    }
  ]
}]

// First call: full price (write to cache)
// Subsequent (within 5 min): 90% off cached portion!

🛠️ Tool Use

tools: [{
  name: "get_weather",
  description: "Get current weather",
  input_schema: {
    type: "object",
    properties: { location: { type: "string" } },
    required: ["location"]
  }
}]

🖥️ Computer Use ⭐

Claude OS/browser control করতে পারে — screenshot দেখে mouse/keyboard action।

tools: [{ type: "computer_20241022", name: "computer",
          display_width_px: 1024, display_height_px: 768 }]
// Screenshot → click(x,y), type, scroll

📁 Files API

PDF, image upload করে context-এ use।

🎯 Claude vs OpenAI

Feature	Claude	OpenAI
Context	200K	128K
Caching	90% off	50% off (batch)
Vision	Yes	Yes
Computer Use	Yes ⭐	Operator
Code	Best-in-class	Excellent
Image gen	No	DALL-E
Voice	No native	TTS, Whisper

💡 When Claude Better

Long document (200K context)
Complex reasoning
Coding task
Repeated system prompt (caching)
Nuanced writing
Following complex instruction

🐍 Python Example

import anthropic

client = anthropic.Anthropic()

message = client.messages.create(
    model="claude-sonnet-4-20250514",
    max_tokens=1024,
    messages=[{"role": "user", "content": "Hello"}]
)
print(message.content[0].text)