@lumetra/node-red-contrib-engram-memory

Node-RED nodes for Engram — durable, explainable memory for AI agents.

Two function-style nodes plus a credentials config node:

• engram store — write a memory to a bucket (POST /v1/buckets/{bucket}/memories)
• engram query — ask a natural-language question against your memories (POST /v1/query)
• engram-config — encrypted credentials node holding your API key and base URL

Install

In your Node-RED editor:

Manage Palette → Install → @lumetra/node-red-contrib-engram-memory

Or from the command line, inside your Node-RED user directory (default ~/.node-red):

npm install @lumetra/node-red-contrib-engram-memory

Restart Node-RED. The nodes appear under the Engram category in the palette.

Configure

1. Create an API key at lumetra.io. Keys look like eng_live_....
2. Drop an engram store or engram query node into a flow.
3. Click the pencil icon next to the Engram field and create a new engram-config entry. Paste the key into API Key. (Stored encrypted by Node-RED.)
4. Set the bucket name. Buckets are auto-created on first write.

Usage

Store

[inject] --> [engram store, bucket=notes] --> [debug]

Set the inject's payload to a string. After deploy + click, the debug panel shows the Engram response with memory_id. msg.engram.memory_id is also populated for convenience.

You can override the bucket per-message by setting msg.bucket.

Query

[inject question] --> [engram query, bucket=notes] --> [debug]

Set msg.question (or msg.payload) to the natural-language question. The output has the full Engram response in msg.payload and the answer string in msg.engram.answer.

Endpoint reference

This package targets the Engram REST API:

| Op | Method + Path | Body | |---|---|---| | Store | POST /v1/buckets/{bucket}/memories | {"content": "..."} | | Query | POST /v1/query | {"query": "...", "bucket": "..."} |

Self-hosted Engram: set Base URL in the config node.

Privacy

See PRIVACY.md.

License

MIT — Lumetra