Claude Chat icon

Claude Chat

Chat with Claude AI using the Agent SDK - supports skills and multi-turn conversations

Claude Chat

AI-generated

Summary

Send a message to the Claude AI model with customizable chat personality, model selection, and options controlling conversation memory and response behavior.

Inputs

  • Chat Mode — Choose the personality style for the conversation, either 'Standard' for a normal helpful assistant or '😄 Humorous' for fun, witty, entertaining responses with jokes and puns.
  • Prompt (required) — The message text to send to Claude.
  • Model — Select the Claude model to use, with options ranging from fast and low cost (Haiku 4.5) to most capable (Opus 4.5). Defaults to the balanced Sonnet 4.5.
  • Options — Additional settings including:
  • System Prompt: custom instructions overriding chat mode defaults
  • Temperature: controls randomness of responses from 0 (deterministic) to 1 (creative)
  • Max Tokens: the maximum tokens allowed in the response
  • Max Budget (USD): the maximum cost allowed for the execution
  • Conversation Mode: boolean to maintain conversation history between executions
  • Enable Skills: boolean to enable Claude Agent SDK skills for enhanced capabilities
  • Permission Mode: how tool permissions are handled ('Default', 'Read Only', 'Accept Edits')

Output shape

a single JSON object containing the AI-generated text under 'output', plus metadata including token usage, estimated cost, session and conversation info, and skills enabled.

The output includes the final textual response from Claude, token usage metrics, estimated cost of the request, session ID (persisted if conversation mode is enabled), conversation turn count, and an array of available skills if enabled. If conversation mode is active, the conversation history is preserved for subsequent calls. Errors are returned as JSON with an 'error' field and success=false if 'Continue On Fail' is enabled. Output text is accessed from 'output' or 'text' fields.

Examples

Example 1: Send a prompt with humorous chat mode and the Opus 4.5 model, limiting max tokens to 500.

Chat Mode: 'humorous', Prompt: 'Tell me a joke about programming', Model: 'opus', Options: {maxTokens: 500}

Discussion