Extract conversation facts and store them in a vector database for retrieval.
npx skills add https://github.com/yuyijiong/paper-to-skill --skill lightmem
Please help me install this Agent Skill. Skill: lightmem Source: https://github.com/yuyijiong/paper-to-skill/tree/main/examples/lightmem Command: npx skills add https://github.com/yuyijiong/paper-to-skill --skill lightmem
LightMem adds persistent memory to LLM-based agents by extracting facts from conversations, storing them in a vector DB, and retrieving relevant memories to inform future interactions.
Install LightMem and initialize LightMemory with a configuration to start capturing and querying memories.
High-intent search queries and answers about installing and using this skill.
Persistent memory is added to LLM agents by extracting facts from conversations, storing them in a vector database, and semantically retrieving relevant memories to ground future interactions.
Yes, local models are fully supported for fact extraction and memory retrieval. The system features a configurable backend that works with Ollama, vllm, and other local model deployments.
Topic segmentation in agent memory management divides conversations into distinct topics to keep long-running agents coherent. This separation ensures extracted facts are properly contextualized before vector database storage.
Yes, updating existing memories is supported within the vector database. The system manages memory by extracting new facts and updating stored information to maintain accurate contextual grounding for agents.
Building a memory-enabled personal assistant involves extracting user preference facts from conversations and storing them in a vector database. Semantic retrieval then injects these relevant memories during generation.
xuiltulAutomate local LLM text generation and chat via Ollama/vLLM.
nirholasCreate and manage vector embeddings for semantic search across multiple providers.
arunodaSearch extracted facts locally using Ollama embeddings and sqlite-vec.
bychrisrStore and retrieve agent knowledge in markdown and yaml files using PARA.
