Use AI memory and context engineering, built by cognee directly in your n8n workflows.

This community node lets you:

• Add text data to a your cognee
• Turn data into AI memory with cognify to apply best context engineering practices
• Run search over your AI memory datasets

n8n is a fair-code licensed workflow automation platform.

• Installation
• Credentials
• Operations
• Usage examples
• Compatibility
• Resources
• Version history
• License

Install from within n8n:

1. In n8n, go to Settings → Community Nodes
2. Click Install and search for n8n-nodes-cognee, or paste the package name directly
3. Confirm the installation

Or install in your n8n instance directory:

npm install n8n-nodes-cognee

Restart n8n after installation if required.

Get your Cognee API key from https://platform.cognee.ai/

Create credentials of type Cognee API in n8n. The node uses these values to authenticate every request:

• Base URL: The base URL of your Cognee API instance. Default in the credential form is https://cognee--cognee-saas-backend-serve.modal.run.
• API Key: Your Cognee API key, sent via X-Api-Key header.

Reference: Cognee API docs at https://cognee--cognee-saas-backend-serve.modal.run/docs.

The node exposes three resources. Each operation maps to a Cognee API endpoint and request body.

• Operation: Add
• Endpoint: POST /api/add
• Fields:
  • Dataset Name (datasetName, required): Name of the Cognee dataset to add text to
  • Text Data (textData, required, multiple): Array of strings to store

Example body sent by the node:

{
  "datasetName": "support_docs",
  "text_data": [
    "FAQ: Reset password via account settings.",
    "Guide: Export data as CSV from dashboard."
  ]
}

• Operation: Cognify
• Endpoint: POST /api/cognify
• Fields:
  • Datasets (datasets, required, multiple): One or more dataset names to cognify
  • Run in Background (runInBackground, optional boolean): If true, returns immediately

Example body sent by the node:

{
  "datasets": ["support_docs"],
  "run_in_background": false
}

• Operation: Search
• Endpoint: POST /api/search
• Fields:
  • Search Type (searchType): One of GRAPH_COMPLETION, GRAPH_COMPLETION_COT, RAG_COMPLETION
  • Datasets (datasets, required, multiple)
  • Query (query, required)
  • Top K (topK, optional number): Defaults to 10

Example body sent by the node:

{
  "searchType": "GRAPH_COMPLETION",
  "datasets": ["support_docs"],
  "query": "How do I export my data?",
  "topK": 5
}

End-to-end example workflow:

1. Add Data (Cognee)
   • Resource: Add Data → Operation: Add
   • Dataset Name: support_docs
   • Text Data: Add one or more strings with your content
2. Cognify (Cognee)
   • Resource: Cognify → Operation: Cognify
   • Datasets: support_docs
   • Run in Background: false (set true for fire-and-forget)
3. Search (Cognee)
   • Resource: Search → Operation: Search
   • Search Type: GRAPH_COMPLETION
   • Datasets: support_docs
   • Query: Your question, e.g. "How do I export my data?"
   • Top K: 5

Tip: If you run Cognify in the background, add a delay or polling step before Search, depending on your dataset size.

Troubleshooting:

• 401/403 errors: Check the API key and that X-Api-Key is accepted by your Cognee instance.
• Connection errors: Verify Base URL and network access from your n8n host.

• Node.js: >= 20.15
• n8n Nodes API: v1

The node depends on n8n-workflow at runtime (peer dependency). It should work on current n8n releases supporting community nodes.

• Cognee API docs: https://cognee--cognee-saas-backend-serve.modal.run/docs
• Package homepage: https://github.com/topoteretes/cognee-n8n

• 0.1.0: Initial release with Add Data, Cognify, and Search operations.

MIT