lily-memory-5-0-0
von ksemaj
Coding Agents & IDEsKurzbeschreibung
Persistent memory plugin for OpenClaw agents.
Beschreibung
Was dieser Skill macht
Persistent memory plugin for OpenClaw agents that survives session resets and restarts. Uses hybrid SQLite FTS5 keyword search combined with Ollama vector embeddings for semantic recall. Automatically extracts and injects relevant facts around each LLM turn.
Combines keyword and semantic search in a single plugin with zero npm dependencies, so agents get accurate retrieval without a separate vector database or additional packages.
Wann du ihn einsetzt
- Remembering user preferences across separate agent sessions
- Tracking project decisions without re-explaining context each time
- Detecting when an agent is stuck repeating the same topic
- Looking up all stored facts about a specific entity on demand
- Running long-term agents that accumulate knowledge over days or weeks
Beispiel-Workflow
User asks: 'What database did we decide to use for the auth service?'
- 1
Extract keywords from the user message
- 2
Run FTS5 keyword search against SQLite memory store
- 3
Run Ollama vector similarity search if enabled
- 4
Merge and deduplicate results from both searches
- 5
Inject matching memories as context before sending to LLM
Agent responds with the previously stored decision: 'You decided on PostgreSQL for the auth service on 2025-11-03, preferring it over MySQL for its JSON support.'
Voraussetzungen
- Ollama with nomic-embed-text model (optional, for semantic search)
Installation
clawhub install ksemaj/lily-memory-5-0-0Führe den CLI-Befehl aus oder füge den Prompt in Claude Code, Cursor, Codex oder deinen Coding-Agenten ein.