Problem
Flowise currently has no self-hosted, decay-weighted memory backend. The available external memory options (Redis, MongoDB, DynamoDB) store raw chat history without semantic relevance ranking. Users running Flowise in air-gapped or GDPR-restricted environments have no way to provide agents with cross-session semantic recall without sending data to an external cloud API.
Proposed Solution
Add a Dakera Memory node to the Memory category — a self-hosted vector memory backend with decay-weighted semantic search.
Dakera (dakera.ai) is an open-source, self-hosted memory server that exposes:
POST /v1/memories — store a memory
POST /v1/memories/search — semantic recall with decay weighting (recent + frequently-accessed memories rank higher)
DELETE /v1/memories — clear memories for a session
Implementation pattern
Follows the established Mem0/Zep node pattern exactly:
class DakeraMemoryExtended extends BaseChatMemory implements MemoryMethods {
// getChatMessages — semantic recall via Dakera + DB message fetch (DESC, reversed)
// addChatMessages — stores user and AI messages independently
// clearChatMessages — deletes Dakera memories + DB chat messages
}
No new npm dependency — uses native fetch to call the Dakera REST API directly, keeping the PR self-contained.
Files
| File |
Purpose |
packages/components/nodes/memory/DakeraMemory/DakeraMemory.ts |
INode + DakeraMemoryExtended implementing MemoryMethods |
packages/components/credentials/DakeraMemoryApi.credential.ts |
password-type credential for API key |
packages/components/nodes/memory/DakeraMemory/dakera.svg |
Node icon |
Self-hosting
docker run -d -p 3000:3000 -e DAKERA_API_KEY=your-key dakera/dakera:latest
No signup, no external API. Works with any Flowise deployment.
Why this matters
Dakera fills a gap the existing memory nodes don't cover:
|
Redis |
MongoDB |
Dakera |
| Data residency |
Your infra |
Your infra |
Your infra |
| Semantic search |
No |
No |
Yes |
| Decay weighting |
No |
No |
Yes |
| Cross-session recall |
Session-scoped |
Session-scoped |
Yes |
I'm happy to submit a PR following Flowise contribution guidelines.
Problem
Flowise currently has no self-hosted, decay-weighted memory backend. The available external memory options (Redis, MongoDB, DynamoDB) store raw chat history without semantic relevance ranking. Users running Flowise in air-gapped or GDPR-restricted environments have no way to provide agents with cross-session semantic recall without sending data to an external cloud API.
Proposed Solution
Add a Dakera Memory node to the
Memorycategory — a self-hosted vector memory backend with decay-weighted semantic search.Dakera (dakera.ai) is an open-source, self-hosted memory server that exposes:
POST /v1/memories— store a memoryPOST /v1/memories/search— semantic recall with decay weighting (recent + frequently-accessed memories rank higher)DELETE /v1/memories— clear memories for a sessionImplementation pattern
Follows the established Mem0/Zep node pattern exactly:
No new npm dependency — uses native
fetchto call the Dakera REST API directly, keeping the PR self-contained.Files
packages/components/nodes/memory/DakeraMemory/DakeraMemory.tspackages/components/credentials/DakeraMemoryApi.credential.tspassword-type credential for API keypackages/components/nodes/memory/DakeraMemory/dakera.svgSelf-hosting
No signup, no external API. Works with any Flowise deployment.
Why this matters
Dakera fills a gap the existing memory nodes don't cover:
I'm happy to submit a PR following Flowise contribution guidelines.