Claude Code (Credentials) icon

Claude Code (Credentials)

Use Claude Code SDK to execute AI-powered coding tasks with authentication from n8n credentials

Actions2

Query

AI-generated

Summary

Send a prompt to Claude Code to start a new AI-assisted coding conversation and receive structured or textual responses based on your query parameters.

Inputs

  • Prompt (required) — The prompt to send to Claude Code, specifying what coding task or question you want to address.
  • Model — Choice of Claude model to use, either 'Sonnet' (fast and efficient) or 'Opus' (most capable). Defaults to 'Sonnet'.
  • Max Turns — Maximum number of conversation turns allowed in this interaction. Defaults to 25.
  • Timeout — Maximum time in seconds to wait before aborting the query. Defaults to 300 seconds.
  • Project Path — Working directory path for Claude Code operations, affecting file system context.
  • Output Format — Format of the output response: 'Structured' for full structured output, 'Messages' for raw messages array, or 'Text' for only the final text result. Defaults to 'Structured'.
  • Allowed Tools — List of tools Claude Code is permitted to use during the conversation, e.g., WebFetch, TodoWrite.
  • Disallowed Tools — List of tools explicitly blocked from use, which take precedence over allowed tools.
  • Additional Options — Additional settings including Debug Mode (enables debug logging), Permission Mode (how permission requests are handled), and System Prompt (extra instructions for Claude).

Output shape

a single JSON object per input item

The output structure depends on the selected Output Format parameter: 'structured' includes detailed messages, summary, metrics, success flag, and session ID; 'messages' includes the raw array of interaction messages; 'text' returns only final result text along with success flag, duration, cost, and session ID. Sensitive information and secrets are redacted from outputs. Errors or timeouts emit error details if 'Continue on Fail' is enabled.

Links

Discussion