engram-memory

n8n community node for Engram by Lumetra — durable, explainable memory for AI agents.

Package Information

Released: 4 mo ago
Downloads: 14 weekly / 75 monthly
Latest Version: 0.1.1
Author: Lumetra

Available Nodes

Engram icon
Engram
Durable, explainable memory for AI agents (Lumetra Engram)

Documentation

n8n-nodes-engram-memory

n8n community node for Engram — durable, explainable memory for AI agents, by Lumetra.

This node lets you store, query, and manage memories in Engram from any n8n workflow. It exposes six operations against the Engram REST API:

Operation What it does
Store Memory Save a fact or piece of context to a bucket
Query Memory Ask a natural-language question and get a synthesized answer with cited memories
List Memories List raw memories in a bucket (paginated)
List Buckets List buckets in your tenant
Delete Memory Delete a single memory by ID
Clear Memories Wipe every memory in a bucket (destructive)

The node is usableAsTool, so you can wire it directly into n8n's AI Agent nodes as a tool.

Install

In your n8n instance:

  1. Go to Settings → Community Nodes.
  2. Click Install a community node.
  3. Enter @lumetra/n8n-nodes-engram-memory and confirm.
  4. After install, search for Engram in the node panel.

Community nodes must be enabled on your n8n instance. For self-hosted n8n, set N8N_COMMUNITY_PACKAGES_ENABLED=true. n8n Cloud users on supported plans can install community nodes from the UI directly.

Credentials

Create an Engram API credential:

  • API Key — your Engram key (starts with eng_live_…). Create one at https://lumetra.io.
  • Base URL — leave at https://api.lumetra.io for Lumetra Cloud, or override for self-hosted Engram.

The credential's Test button calls GET /v1/buckets?limit=1 to verify the key.

Example workflow

A minimal "remember then recall" workflow:

  1. Engram → Store Memory with bucket = chatbot, content = "Acme's primary contact is Dana Lee, COO."
  2. Engram → Query Memory with bucket = chatbot, question = "Who's the contact at Acme?"
  3. The query node returns { success, answer, memories_found, explanation: { retrieved_memories: [...] } }.

For multi-agent setups, give each project its own bucket name.

Field mapping note

The user-facing Question field maps to the REST API's query field — we use "question" here to stay consistent with Engram's other surfaces (Dify, Claude, ChatGPT plugins).

Development

npm install
npm run build   # tsc → dist/

To test locally inside Docker n8n:

docker run -it --rm \
  -p 5678:5678 \
  -e N8N_COMMUNITY_PACKAGES_ENABLED=true \
  -v ~/.n8n:/home/node/.n8n \
  docker.n8n.io/n8nio/n8n

Then Settings → Community Nodes → Install and enter @lumetra/n8n-nodes-engram-memory (after the package is published to npm), or use npm link for an unpublished local build.

Privacy

See PRIVACY.md.

License

MIT © Lumetra, Inc.

Discussion