메뉴
n8n node reference — core nodes, app/action nodes, AI cluster nodes (agents, LLMs, memory, tools, vector stores). Node parameters, operations, and doc links. Auto-load on questions about specific n8n node types, parameters, AI agent setup, LLM nodes, RAG chains, or "which node should I use for X".
Codex 또는 Claude로 설치 이 Prompt를 복사해 Codex, Claude 또는 다른 어시스턴트에 붙여 넣으면 Skill 페이지를 검토하고 설치를 진행할 수 있습니다.
agent-browser CLI — headless Chromium for AI agents. Snapshot-and-ref workflow, core commands, and yai service interaction patterns. Auto-load when asked to open a browser, navigate a URL, click, fill, screenshot, or interact with any yai web UI (Grafana, n8n, Windmill, Langfuse, MinIO, etc.).
Browserless headless Chromium service on the yai stack — REST endpoints (screenshot, PDF, scrape, function), WebSocket CDP/Puppeteer/Playwright patterns, token auth, sizing. Auto-load on Browserless, headless browser, Puppeteer, Playwright, CDP, screenshot, PDF, scrape-script questions.
Generate clean architecture / flow diagrams as PNG+SVG for the yai stack. Hand-laid-out Python→SVG→PNG pipeline (no Mermaid auto-layout). Auto-load when asked to create, update, or render an architecture diagram, flow chart, system diagram, or "draw the stack".
Firecrawl web crawling & scraping API on the yai stack — scrape, crawl, extract endpoints, job status, concurrency knobs, proxy config. Auto-load on Firecrawl, web scrape, crawl, extract, markdown, web content questions.
Grafana on the yai stack — datasource UIDs, file provisioning (not TF), dashboard conventions, scrape targets page, Grafana plugin setup. Auto-load on Grafana, dashboard, datasource, DS_*, panel, templating variable, provisioning questions.
How to control every yai web UI — login recipes, navigation patterns, and common actions for Grafana, Langfuse, MinIO, n8n, Windmill, LiteLLM, Qdrant, and Traefik. Auto-load when asked to log in, navigate, click, or perform any action in a yai service UI.
| name | n8n-docs |
| description | n8n node reference — core nodes, app/action nodes, AI cluster nodes (agents, LLMs, memory, tools, vector stores). Node parameters, operations, and doc links. Auto-load on questions about specific n8n node types, parameters, AI agent setup, LLM nodes, RAG chains, or "which node should I use for X". |
| when_to_use | Load when the user asks which n8n node to use for a specific task, how to configure a node's parameters, or how to build an AI agent, RAG chain, or LLM workflow in n8n. Also load when troubleshooting a specific node type's behavior. |
Official docs root: https://docs.n8n.io/integrations/builtin/node-types/
Full list: https://docs.n8n.io/integrations/builtin/core-nodes/
| Node | Doc path | Notes |
|---|---|---|
| Manual Trigger | core-nodes/n8n-nodes-base.manualTrigger | Start by hand |
| Schedule Trigger | core-nodes/n8n-nodes-base.scheduleTrigger | Cron or interval |
| Webhook | core-nodes/n8n-nodes-base.webhook | HTTP POST/GET receiver |
| Email Trigger (IMAP) | core-nodes/n8n-nodes-base.emailimap | Trigger on new email |
| RSS Feed Trigger | core-nodes/n8n-nodes-base.rssFeedReadTrigger | New RSS item |
| Chat Trigger | core-nodes/n8n-nodes-base.chatTrigger | Used with AI Chat nodes |
| Local File Trigger | core-nodes/n8n-nodes-base.localFileTrigger | File system events |
| SSE Trigger | core-nodes/n8n-nodes-base.sseTrigger | Server-sent events |
| n8n Form Trigger | core-nodes/n8n-nodes-base.formTrigger | Web form submissions |
| Execute Sub-workflow Trigger | core-nodes/n8n-nodes-base.executeWorkflowTrigger | Called by another workflow |
| Error Trigger | core-nodes/n8n-nodes-base.errorTrigger | Catch workflow errors |
| Node | Doc path | Notes |
|---|---|---|
| If | core-nodes/n8n-nodes-base.if | Boolean branch; use typeValidation: "loose" for mixed types |
| Switch | core-nodes/n8n-nodes-base.switch | Multi-branch routing |
| Loop Over Items | core-nodes/n8n-nodes-base.splitInBatches | Batch iterator; typeVersion 3: output 1 = items, output 0 = done |
| Wait | core-nodes/n8n-nodes-base.wait | Pause execution (time or webhook resume) |
| Stop And Error | core-nodes/n8n-nodes-base.stopAndError | Halt with custom error message |
| Execute Sub-workflow | core-nodes/n8n-nodes-base.executeWorkflow | Call another workflow |
| No Operation | core-nodes/n8n-nodes-base.noOp | Placeholder / passthrough |
| Node | Doc path | Notes |
|---|---|---|
| Edit Fields (Set) | core-nodes/n8n-nodes-base.set | Add/overwrite/rename fields |
| Filter | core-nodes/n8n-nodes-base.filter | Remove items by condition |
| Sort | core-nodes/n8n-nodes-base.sort | Order items |
| Merge | core-nodes/n8n-nodes-base.merge | Combine two inputs |
| Split Out | core-nodes/n8n-nodes-base.splitOut | Flatten array field to items |
| Remove Duplicates | core-nodes/n8n-nodes-base.removeDuplicates | Dedup by field(s) |
| Rename Keys | core-nodes/n8n-nodes-base.renameKeys | Rename field keys |
| Aggregate | core-nodes/n8n-nodes-base.aggregate | Group items, collect to array |
| Compare Datasets | core-nodes/n8n-nodes-base.compareDatasets | Diff two item sets |
| Limit | core-nodes/n8n-nodes-base.limit | Cap item count |
| Summarize | core-nodes/n8n-nodes-base.summarize | Group + aggregate (count, sum, avg) |
| Data Table | core-nodes/n8n-nodes-base.dataTable | Tabular ops |
| Node | Doc path | Notes |
|---|---|---|
| HTTP Request | core-nodes/n8n-nodes-base.httpRequest | Generic REST/API calls |
| GraphQL | core-nodes/n8n-nodes-base.graphql | GraphQL queries |
| Respond to Webhook | core-nodes/n8n-nodes-base.respondToWebhook | Send HTTP response in webhook workflows |
| RSS Read | core-nodes/n8n-nodes-base.rssFeedRead | Fetch RSS feed once |
| Node | Doc path | Notes |
|---|---|---|
| Code | core-nodes/n8n-nodes-base.code | JS or Python; runs in external runner |
| Execute Command | core-nodes/n8n-nodes-base.executeCommand | Shell command on host |
| LangChain Code | cluster-nodes/root-nodes/n8n-nodes-langchain.code | Custom LangChain JS |
| Node | Doc path | Notes |
|---|---|---|
| Read/Write Files | core-nodes/n8n-nodes-base.readWriteFile | Host filesystem |
| Extract From File | core-nodes/n8n-nodes-base.extractFromFile | CSV, JSON, XML, binary → items |
| Convert to File | core-nodes/n8n-nodes-base.convertToFile | Items → CSV, JSON, XLSX, etc. |
| Compression | core-nodes/n8n-nodes-base.compression | zip/gzip/tar |
| HTML | core-nodes/n8n-nodes-base.html | Parse / generate HTML |
| XML | core-nodes/n8n-nodes-base.xml | Parse / generate XML |
| Markdown | core-nodes/n8n-nodes-base.markdown | MD ↔ HTML conversion |
| Edit Image | core-nodes/n8n-nodes-base.editImage | Resize, crop, annotate images |
| Node | Doc path | Notes |
|---|---|---|
| Send Email | core-nodes/n8n-nodes-base.sendEmail | SMTP |
| n8n Form | core-nodes/n8n-nodes-base.form | Render interactive form step |
| Node | Doc path | Notes |
|---|---|---|
| Crypto | core-nodes/n8n-nodes-base.crypto | Hash, HMAC, sign, encrypt |
| JWT | core-nodes/n8n-nodes-base.jwt | Sign and verify JWTs |
| TOTP | core-nodes/n8n-nodes-base.totp | Time-based OTP |
| Node | Doc path | Notes |
|---|---|---|
| FTP | core-nodes/n8n-nodes-base.ftp | FTP/SFTP file ops |
| SSH | core-nodes/n8n-nodes-base.ssh | Remote shell commands |
| Git | core-nodes/n8n-nodes-base.git | Git repo operations |
| LDAP | core-nodes/n8n-nodes-base.ldap | Directory queries |
| MQTT | (app node) | Pub/sub messaging |
| AMQP Sender | (app node) | RabbitMQ / AMQP |
| Node | Doc path | Notes |
|---|---|---|
| AI Transform | core-nodes/n8n-nodes-base.aiTransform | Apply LLM to each item without building a chain |
| MCP Client | core-nodes/n8n-nodes-base.mcpClient | Call an MCP server endpoint |
| MCP Server Trigger | core-nodes/n8n-nodes-base.mcpServerTrigger | Expose workflow as MCP tool |
Docs hub: https://docs.n8n.io/advanced-ai/
Node list: https://docs.n8n.io/integrations/builtin/cluster-nodes/
Cluster nodes consist of a root node + one or more sub-nodes wired to its side inputs.
| Node | Purpose | Doc |
|---|---|---|
| AI Agent | Autonomous agent — selects tools, iterates until goal met | https://docs.n8n.io/integrations/builtin/cluster-nodes/root-nodes/n8n-nodes-langchain.agent/ |
| Basic LLM Chain | Simple prompt → LLM → output; no tools | https://docs.n8n.io/integrations/builtin/cluster-nodes/root-nodes/n8n-nodes-langchain.chainllm/ |
| Question and Answer Chain | RAG: retrieve from vector store, answer with LLM | — |
| Summarization Chain | Map-reduce or stuff summarization of long docs | — |
| Information Extractor | Extract structured JSON from unstructured text | — |
| Text Classifier | Classify text into predefined categories | — |
| Sentiment Analysis | Detect positive / negative / neutral sentiment | — |
| LangChain Code | Drop-in custom LangChain JS chain | — |
| Vector Store (insert) | Upsert embeddings into a vector DB | — |
| Vector Store (retrieve) | Similarity search + optional Q&A | — |
| Type | When to use |
|---|---|
| Tools Agent (default) | Best all-around; enforces structured output; use with any LLM that supports tool-calling |
| Conversational Agent | Dialogue flows; keeps conversation history |
| ReAct Agent | Complex multi-step reasoning; Reason + Act loops |
| Plan and Execute Agent | Decomposes task into plan first, then executes steps |
| OpenAI Functions Agent | Uses OpenAI function-calling API specifically |
| SQL Agent | Generates and runs SQL against a DB; auto-discovers schema |
Attach to root node's Language Model input.
| Sub-node | Provider | Notes |
|---|---|---|
lmChatOpenAi | OpenAI | GPT-4o, GPT-4.1, o3, etc. For LiteLLM: base URL = http://host.docker.internal:24000/v1 |
lmChatAnthropic | Anthropic | Claude 3.x / Claude 4.x |
lmChatGoogleGemini | Google Gemini | |
lmChatMistralCloud | Mistral AI | |
lmChatGroq | Groq | Fast inference |
lmChatAwsBedrock | AWS Bedrock | Claude, Llama, Titan |
lmChatAzureOpenAi | Azure OpenAI | |
lmChatOllama | Ollama | Local models (not in yai — no GPU) |
lmChatOpenRouter | OpenRouter | Multi-provider routing |
lmChatDeepSeek | DeepSeek | |
lmChatXAI | xAI Grok | |
lmChatMiniMax | MiniMax | |
lmChatMoonshot | Moonshot Kimi | |
lmChatAlibabaTongyi | Alibaba |
Attach to root node's Embeddings input (vector store insert/retrieve).
| Sub-node | Provider |
|---|---|
embeddingsOpenAi | OpenAI (text-embedding-3-small default) |
embeddingsGoogleGemini | |
embeddingsCohere | Cohere |
embeddingsAWSBedrock | AWS Bedrock |
embeddingsAzureOpenAi | Azure OpenAI |
embeddingsHuggingFace | HuggingFace Inference |
embeddingsMistralCloud | Mistral |
embeddingsOllama | Ollama (local) |
Attach to root node's Memory input. Persists conversation context across agent turns.
| Sub-node | Backend | Notes |
|---|---|---|
memoryBufferWindow | In-memory | Simple sliding window; lost on restart |
memoryChatManager | Configurable | Chat Memory Manager — explicit get/set |
memoryPostgresChat | Postgres | Persistent; use yai-postgres |
memoryRedisChat | Redis | Fast; use yai-n8n-redis-1 |
memoryMongoDbChat | MongoDB | |
memoryMotorhead | Motorhead server | |
memoryXata | Xata DB | |
memoryZep | Zep |
Attach to root node's Tools input. Agents call these to act on the world.
| Sub-node | What it does |
|---|---|
toolAiAgent | Delegate to a child AI Agent (multi-agent orchestration) |
toolMcpClient | Expose any MCP server's tools to the agent |
toolWorkflow | Call another n8n workflow as a tool |
toolCode | Arbitrary JS function the agent can invoke |
toolCalculator | Math expressions |
toolWikipedia | Wikipedia search |
toolSerpApi | Google search via SerpApi key |
toolSearXNG | Self-hosted SearXNG web search |
toolWolframAlpha | Wolfram |
toolVectorStoreSearch | Semantic search against a vector store |
toolThink | Internal reasoning step (Claude extended thinking style) |
Used as both root nodes (insert/retrieve) and as retriever sub-nodes.
| Backend | Sub-node / Root node |
|---|---|
| Qdrant (yai default) | vectorStoreQdrant — host: http://host.docker.internal:26000 |
| PGVector (Postgres) | vectorStorePGVector |
| Chroma | vectorStoreChroma |
| Pinecone | vectorStorePinecone |
| Weaviate | vectorStoreWeaviate |
| Redis | vectorStoreRedis |
| Supabase | vectorStoreSupabase |
| MongoDB Atlas | vectorStoreMongoDBAtlas |
| Milvus | vectorStoreMilvus |
| Azure AI Search | vectorStoreAzureAiSearch |
| Simple (in-memory) | vectorStoreInMemory |
| Zep | vectorStoreZep |
Attach to vector store root node's Document input.
| Sub-node | Source |
|---|---|
documentDefaultDataLoader | n8n binary data (file from previous node) |
documentGithubLoader | GitHub repo files |
Attach to document loaders. Control chunk size for embedding.
| Sub-node | Strategy |
|---|---|
textSplitterCharacter | Split on a delimiter character |
textSplitterRecursiveCharacter | Recursive by paragraph/sentence/word (recommended default) |
textSplitterTokens | Split by token count |
Attach to LLM chain / agent to enforce structured output.
| Sub-node | Output format |
|---|---|
outputParserStructured | JSON Schema → typed object |
outputParserItemList | Comma-separated list → array |
outputParserAutofixing | Wraps another parser + retries on malformed output |
Used in Q&A chains for advanced retrieval strategies.
| Sub-node | Strategy |
|---|---|
retrieverVectorStore | Standard similarity search |
retrieverContextualCompression | Compress retrieved docs with LLM |
retrieverMultiQuery | Generate multiple query variants, merge results |
retrieverWorkflow | Custom retrieval via another n8n workflow |
Reorder retrieved results by relevance.
| Sub-node | Notes |
|---|---|
rerankerCohere | Cohere Rerank API |
rerankerModelSelector | Pick best document with an LLM |
Full list: https://docs.n8n.io/integrations/builtin/app-nodes/
Each app node doc page covers:
| Node | Package name | Notes |
|---|---|---|
| Postgres | n8n-nodes-base.postgres | Use yai internal DSN; avoid $body$ dollar-quoting |
| Slack | n8n-nodes-base.slack | Use messageType: "text" not "block" |
| HTTP Request | n8n-nodes-base.httpRequest | Used for LiteLLM API, MinIO S3, Qdrant REST, etc. |
| OpenAI | n8n-nodes-base.openAi | Can point to LiteLLM base URL |
| Qdrant | (HTTP Request) | No native node; use HTTP Request against http://host.docker.internal:26000 |
// Access current item
$json.fieldName
$json["field-with-dash"]
// Access item from another node
$('NodeName').item.json.field
$('NodeName').all() // array of all items
$('NodeName').first().json // first item
// Workflow / execution metadata
$workflow.id
$execution.id
// Date helpers
$now // current DateTime (Luxon object)
$today // today at midnight
$now.toISO() // ISO string
$now.minus({hours: 6}).toISO()
// Env / secrets (Code node only)
process.env.MY_VAR
| Resource | URL |
|---|---|
| Node types overview | https://docs.n8n.io/integrations/builtin/node-types/ |
| Core nodes | https://docs.n8n.io/integrations/builtin/core-nodes/ |
| App nodes | https://docs.n8n.io/integrations/builtin/app-nodes/ |
| Cluster nodes (AI) | https://docs.n8n.io/integrations/builtin/cluster-nodes/ |
| Advanced AI hub | https://docs.n8n.io/advanced-ai/ |
| LangChain concepts | https://docs.n8n.io/advanced-ai/langchain/langchain-n8n/ |
| AI Agent node | https://docs.n8n.io/integrations/builtin/cluster-nodes/root-nodes/n8n-nodes-langchain.agent/ |
| Tools Agent | https://docs.n8n.io/integrations/builtin/cluster-nodes/root-nodes/n8n-nodes-langchain.agent/tools-agent/ |
| Basic LLM Chain | https://docs.n8n.io/integrations/builtin/cluster-nodes/root-nodes/n8n-nodes-langchain.chainllm/ |
| Community nodes | https://docs.n8n.io/integrations/community-nodes/installation/ |
| Expressions | https://docs.n8n.io/code/expressions/ |
| Workflow JSON schema | https://docs.n8n.io/workflows/export-import/ |