Claude-Powered n8n Automation Blueprints for Repetitive Tasks

Turn “I hate doing this task” into an n8n automation blueprint—powered by Claude and returned as clean JSON

This n8n automation turns a visitor’s message into a structured automation blueprint: an Anthropic Claude agent asks a few focused clarifying questions, then outputs a strict, frontend-ready JSON object—automatically validated and auto-fixed to ensure reliable responses.

What this workflow does

• Receives a POST request on a header-authenticated webhook with { "message": "...", "history": [] } (history is optional).
• Validates input: if message is missing or empty, it returns a 400 JSON error response.
• Builds a Claude prompt that summarizes the conversation, counts prior follow-up questions, and enforces a maximum follow-up limit.
• Calls Anthropic Claude via an agent flow: it either asks one clarifying question or produces a blueprint JSON object.
• Parses and auto-fixes output against a strict structured schema so the result is valid JSON every time.
• Returns a consistent response contract to the webhook caller with caching disabled.

Use cases

• Getting an n8n workflow plan for a repetitive process like collecting requirements, triaging requests, or standardizing outputs.
• Building a conversational “automation assistant” for a SaaS frontend where users need renderable JSON back from the backend.
• Guiding automation engineers through discovery with controlled follow-up questions before generating the blueprint.

Technical details

• Anthropic API using Claude (two Anthropic Chat Model nodes: one for the agent and one for the auto-fix parser).
• Header Auth webhook security for protected access.
• Nodes used: if, set, code, gmail, sticky note, google sheets.

Setup includes adding Anthropic credentials, configuring Header Auth, copying the production webhook URL, and adjusting maxFollowUps plus agent system instructions to match your discovery process.