Merged
Size
XL
Change Breakdown
Feature85%
Maintenance15%
#28189feat(Moonshot Kimi Node): Add new node

Moonshot Kimi node brings AI chat and vision to n8n

Moonshot Kimi node brings AI chat and vision to n8n

n8n workflows can now tap into Moonshot Kimi models for AI chat and image analysis, with built-in tool calling, web search, and thinking mode support.

n8n users gain a dedicated integration for Moonshot's Kimi AI models, expanding the automation platform's AI capabilities beyond existing vendors. The new node handles both conversational text generation and vision tasks through two distinct operations.

The Message a Model resource delivers a full-featured chat interface with multi-turn conversation history, system prompt support, and configurable output parameters like temperature and penalties. Tool calling lets workflows connect to external services or tools, with a configurable iteration limit and cancellation support. A built-in web search toggle activates Kimi's $web_search capability for real-time information retrieval, while thinking mode enables deep reasoning for complex tasks. JSON response formatting is available for structured outputs.

The Analyze Image resource provides a streamlined vision interface that accepts binary image data and returns analysis. Unlike some vision APIs, Moonshot requires base64-encoded data URLs rather than image URLs—a constraint the implementation handles automatically.

Dynamic model selection pulls available models directly from Moonshot's API, letting users pick from the list or enter model IDs manually. Token usage tracking feeds into n8n's telemetry system for monitoring. The node reuses existing Moonshot API credentials, minimizing configuration friction.

The node lives in the LangChain nodes package alongside other AI vendor integrations, available for workflows targeting both text and image processing tasks.

View Original GitHub Description

Summary

Loom Recording

Add a standalone Moonshot Kimi vendor node with two resources:

  • Text > Message a Model - Full-featured chat with system prompts, multi-turn conversation, tool calling, built-in web search ($web_search), thinking mode, JSON response format, and image attachments (binary)
  • Image > Analyze Image - Dedicated vision operation for image analysis via binary input

Key implementation details:

  • Dynamic model selection via /models endpoint (resource locator)
  • Token usage tracking for telemetry (accumulateTokenUsage)
  • Tool calling loop with cancellation support and configurable max iterations
  • Thinking mode toggle with explicit thinking: { type: 'disabled' } sent by default on all requests to prevent models like kimi-k2.5 from silently consuming tokens on reasoning
  • Conditional parameter sending (temperature, penalties) to avoid model-specific restrictions
  • Binary-only image input (Moonshot API does not support URL-formatted images, only base64)
  • Reuses moonshotApi credentials and logo from the chat model sub-node

Related Linear tickets, Github issues, and Community forum posts

https://linear.app/n8n/issue/NODE-4761

Review / Merge checklist

  • PR title and summary are descriptive. (conventions)
  • Docs updated or follow-up ticket created.
  • Tests included.
  • PR Labeled with release/backport (if the PR is an urgent fix that needs to be backported)

Made with Cursor

© 2026 · via Gitpulse