feat: add Stellar Memory filter pipeline

[sangjun0000]

Stellar Memory Filter Pipeline

Adds a filter pipeline that integrates Stellar Memory — a local AI memory system — with OpenWebUI.

What it does

• inlet: Before each request, searches Stellar Memory for relevant context and prepends it to the system prompt automatically
• outlet: After each response, optionally saves the assistant's reply as a new memory (auto-growing knowledge base)

Requirements

• Stellar Memory running locally: npx stellar-memory api (starts REST API on port 21547)
• No cloud service required — fully local and private

Configuration (Valves)

Valve	Default	Description
STM_URL	http://localhost:21547	Stellar Memory API URL
STM_PROJECT	"" (auto: "default")	Project namespace for memory isolation
RECALL_LIMIT	5	Max memories to inject per request
AUTO_REMEMBER	True	Auto-save assistant responses as memories

Setup

1. Install Stellar Memory: npm install -g stellar-memory && npm run init
2. Start the API: npx stellar-memory api
3. Upload this pipeline in OpenWebUI → Admin → Pipelines
4. Set STM_URL in pipeline settings (Docker users: use http://host.docker.internal:21547)