Filesystem-based persistent agent memory: agents read, write and search memory while a background daemon handles consolidation and extraction.
Persistent memory with SQLite hybrid search (FTS5 and semantic), built-in Qwen3 embeddings and rclone sync across machines.
Retrieve context from your Ragie knowledge base, connected to sources like Google Drive, Notion and Jira.
LLM-driven context and memory management: short- and long-term memory with vector, timeline and knowledge graph retrieval plus reranking.
SQLite-based memory with exponential decay: memories strengthen on access and fade when unused, limiting unbounded context growth.
Personal knowledge graph for Claude on Neo4j with typed entities, bi-temporal edges, weighted decay, contradiction detection and local embeddings.
STIL lab at ETS Montreal — knowledge base and paper search via ArXiv and Semantic Scholar.
Free MCP: relationship_memory validate, store (local), retrieve — portable interlocutor cards v1.
Local-first memory for coding agents that turns coding sessions into reusable engineering memory: decisions, gotchas and domain knowledge.
Persistent memory with swappable Postgres/pgvector, Qdrant or Neo4j backends, adaptive retrieval weighting, duplicate merging and stale-fact archiving.
Production-ready RAG + MCP demo: eval-in-CI merge gate, Langfuse traces, structure-aware chunking.
Long-term memory for AI coding agents, stored as a tree of plain local files.
Agent memory MCP server with provenance tracking, decay-weighted recall, and feedback loops.
Cloud-hosted MCP server for durable AI memory
Structural Thinking Engine: MCP server integrating Gemini CLI with authentic creative processes, structural observation protocols, and persistent memory charts based on Robert Fritz's methodology
Local-first long-term memory for AI agents, with byte-recomputable signed verification receipts.
A personality, mood and memory layer for AI agents on an exact p-bit decision engine: an AI persona outside the model, every stance computed on a CPU, rules kept by construction, replayable and verifi
Give your AI a soul — memories, corrections, daily logs and a knowledge base that persist across Claude, Cursor, Windsurf and VS Code. Imports your AI-tool history into a local-first brain.
Private on-device memory & retrieval for AI assistants — offline vector + lexical search.
MCP server for the OpenAI ChatGPT API — chat, vision, and embeddings
Governance-grade knowledge-base format (OKF-compatible) plus CLI and single-writer MCP server
Local visual UI cache for AI agents using perceptual hashing, CLIP, and 3D spatial grounding.
Shared memory for AI agents, as a graph in your own Postgres. Writes never call an LLM.
Local-first knowledge management with bi-directional LLM sync via Markdown files.
Claude Code plugin for llmkb — search, read, and write your knowledge base via MCP tools
MCP server providing semantic code search and API reference for the Immersive Web SDK using RAG (Retrieval-Augmented Generation) with vector embeddings
Local RAG search over Obsidian vaults: SQLite hybrid search with local Ollama embeddings.
Local-first memory for AI agents, over MCP on your machine. Agents propose; only you confirm.
LLM & RAG evals: hallucination, eval contracts, semantic tests, tokens & model info. Free, no key.
Compresses AI chat sessions into a typed knowledge graph. Local/offline via Ollama.
General-purpose Markdown memory with collections, tags, text and semantic search over MCP and CLI.
Fresh RSS/Atom items, published podcast transcripts and public PDF text via Apify.
Cursor plugin for HarnMem, the cross-harness agent memory: byte-stable first-prompt injection, five MCP memory tools, and session-end journaling. One shared store of plain, human-editable Markdown und
Primitive RAG-like web search that runs locally, without external APIs.
Structural codebase MCP server with persistent memory: navigate by symbol, recall across sessions.
Reptile and exotic pet knowledge base: species identification, genetics calculator, care information, health diagnosis, prices and blockchain provenance.
ไดเรกทอรีทั้งหมด
RAG กราฟ และหน่วยความจำระยะยาวของเอเจนต์ 2,590 เซิร์ฟเวอร์ เรียงจากที่ใช้มากที่สุด แต่ละรายการจะเปิดหน้าของผู้เผยแพร่