n8n LlamaParse Platform node

This repository contains a custom n8n community node for integrating with the LlamaParse Platform, bringing document parsing, classification, splitting, structured extraction, and index retrieval into your n8n workflows.

The package ships a single node — LlamaParse Platform — exposing five resources that map to the LlamaParse Platform APIs.

🚀 Resources

The LlamaParse Platform node lets you pick a resource from a dropdown, then choose an operation. All resources share the same LlamaParse API credential.

📄 Parse

Parse documents (PDF, DOCX, PPTX, images, etc.) into clean text or Markdown. Supports four tiers — fast, cost_effective, agentic, and agentic_plus — and pinned parser versions.

• Input: binary file (from any previous node)
• Output: { text: string } (Markdown when available, otherwise plain text)

🏷️ Classify

Route a document into one of a set of user-defined categories. You provide a list of { category, description } rules and get back the predicted category along with confidence and reasoning.

• Input: binary file + classification rules
• Output: { category, reasons, confidence }

✂️ Split

Split a multi-section document into segments using semantic categories you define (e.g. "invoice", "contract", "appendix"). You can choose how uncategorized content is handled (include / forbid / omit).

• Input: binary file + categories
• Output: one item per segment, with { category, confidence, pages }

⚗️ Extract

Run a pre-configured LlamaExtract agent against a file to get back structured JSON matching your schema.

• Input: binary file + Extraction Configuration ID
• Output: { result: <stringified JSON> }

🔎 Retrieve (Index v1 · Index v2)

Query a LlamaCloud index/pipeline and return the top-k retrieved chunks. Works against both the v1 Pipelines API and the newer v2 Index Retrieval API — ideal for RAG flows with the n8n AI Agent / Chat Trigger nodes.

• Input: indexId, query (defaults to {{ $json.chatInput }}), topK
• Output: { context: string[] }

The node is also marked usableAsTool: true, so any of the resources above can be wired into an n8n AI Agent as a tool.

📋 Prerequisites

1. Node.js ≥ 20 — install via nvm on Linux/macOS/WSL or the Microsoft guide on Windows.
2. n8n installed globally:

   npm install n8n -g

3. A LlamaParse Platform API key — sign up at cloud.llamaindex.ai and copy the key from the dashboard.

🛠️ Installation

Option 1 — Install from npm (recommended)

npm install -g @llamaindex/n8n-nodes-llamacloud

mkdir -p ~/.n8n/custom
cd ~/.n8n/custom
npm link @llamaindex/n8n-nodes-llamacloud

n8n start

Option 2 — Local development from source

git clone https://github.com/run-llama/n8n-llamacloud.git
cd n8n-llamacloud
npm install
npm run build
npm link

mkdir -p ~/.n8n/custom
cd ~/.n8n/custom
npm link @llamaindex/n8n-nodes-llamacloud

n8n start

Option 3 — Docker

A Dockerfile and compose.yaml are provided at the repository root. See the Docker guide for the full walkthrough.

🔧 Configuring credentials

1. Open n8n (default: http://localhost:5678).
2. Go to Credentials → Add Credential → search for "LlamaParse API".
3. Paste your API key. Optionally override the base URL (defaults to https://api.cloud.llamaindex.ai).
4. Save.

Full screenshots: guides/index.md.

📚 Guides

Step-by-step walkthroughs (with screenshots) for every resource live under guides/:

• Setup & credentials
• Parse
• Classify
• Split
• Extract
• Retrieve — Index v1
• Retrieve — Index v2
• Run with Docker

🔗 References

• LlamaParse Platform
• LlamaCloud documentation
• n8n documentation
• n8n community

📝 License

MIT