#retrieval

Mnemosyne OS 8.0.0 — zero-dependency, local-first AI memory system (MCP / API / CLI / Python). MIT. https://ai-memory.net

AccInt - local-first MCP Work Model for coding agents that learns from real outcomes.

Relation-aware, calibrated and provenance-preserving context routing for interleaved agent conversations

Nokori (残り) — behavioral memory layer for Claude Code and Cursor. Turns your corrections into rules that come back when it counts. 经验留下的痕迹，比记忆更深。

A personal memory layer for your AI tools. Intelligent authoring (gap detection, self-enriching graph, session-synthesis loop) + measurably good retrieval (94.9% blind Cov@10, sub-ms latency). Domain-agnostic framework — works for any context you want to track. MCP-native, Claude-Code-first, works with Cursor/Continue/Cline/OpenAI/Gemini.

AI research chatbot with dynamic query planning, parallel search, and RAGAS-evaluated grounded answers.

Java RAG knowledge base with controlled Multi-Agent repair: Tool Calling, HITL approval, versioned rollback, pgvector retrieval and reproducible evals.

多模态内容理解与工具编排 Agent：Chinese-CLIP 图文共享向量空间 + BM25/RRF 混合检索 + cross-encoder 重排（真实文本查询 hit@10 0.668→0.918）+ Schema 约束工具调用 + 可溯源生成，含 MCP、多轮会话记忆、ReAct 编排与分层离线评测。 | Multimodal agent: Chinese-CLIP retrieval, BM25+RRF hybrid search, cross-encoder reranking, schema-constrained tool calling, MCP server.

Apex-relevance subgraph retrieval for AI agents — feed your LLM the most relevant slice of a knowledge graph, within a token budget.

A context database for LLMs: structured storage and hybrid retrieval (BM25 + vectors, RRF-fused, token-budgeted) over SQLite, exposed to Claude as an MCP server

Your agent will answer with nothing behind it. This tells you how often. Measures whether an agent can actually reach the right doc in your corpus.

A searchable, self-maintaining catalog over any corpus of tools, skills, agents, and components.