// Prompt Caching Example
// Demonstrates how to enable caching on system prompts and tools
// Reduces API costs by ~90% on repeated requests with cached content

claude = require("claude")

// Define a set of tools that will be cached
var tools = [
  {
    name = "search_docs",
    description = "Search the knowledge base documentation",
    parameters = {
      query = {type = "string", description = "Search query"},
      max_results = {type = "number", description = "Max results to return"}
    },
    required = ["query"],
    handler = function(input)
      // Simulated search results
      return "Found 5 documents matching: " + input.query
    end
  },
  {
    name = "get_version",
    description = "Get the current version",
    parameters = {},
    required = [],
    handler = function(input)
      return "v0.21.10"
    end
  }
]

// System prompt - same for all requests in this session
var system_prompt = """
You are a helpful documentation assistant. You have access to tools to search
the knowledge base and retrieve system information. Always be accurate and cite
your sources when providing information.

Your goal is to help users find information quickly and effectively.
"""

// Create a session with caching enabled
// This is perfect for scenarios where you'll be asking multiple questions
// with the same system prompt and tools - like a support chatbot
var chat = claude.session({
  system = system_prompt,
  tools = tools,
  cache_control = {
    system = true,  // Cache the system prompt
    tools = true    // Cache all tool definitions
  }
})

print("Documentation Assistant Ready")
print("=" * 40)
print("")
print("System prompt and tools are cached.")
print("First request builds cache (normal token cost)")
print("Subsequent requests reuse cache (~90% savings on cached tokens)")
print("")

// Example multi-turn conversation
// Note: This example shows the structure, but won't make actual API calls
// To use with real Claude API, set ANTHROPIC_API_KEY environment variable

var question1 = "What is Duso?"
print("Q: " + question1)
// response1 = chat.prompt(question1)
// print("A: " + response1)
// print("Usage so far:", format_json(chat.usage))
// print("")

var question2 = "Tell me more about its features"
print("Q: " + question2)
// response2 = chat.prompt(question2)
// print("A: " + response2)
// print("Final usage:", format_json(chat.usage))
// print("Note: Second response reused cached system prompt & tools!")

print("")
print("Cache Benefits:")
print("- Faster responses (server-side caching)")
print("- Lower costs (90% reduction on cached content)")
print("- Better for multi-turn conversations")
print("- Especially valuable with large system prompts or many tools")