一个基于 Arcade 的 AI 智能体个人知识库，适用于模型上下文协议（MCP）。

**状态。**代理库是由 Arcade.dev 工程师开发的社区项目。它不是官方 Arcade.dev 产品，不受 Arcade.dev 支持或SLA覆盖。请参见下方支持，了解预期情况。

命名。 包是 Agent Library（在 PyPI 上用 [agent-library，librarian 作为 Python 模块）。其中的 MCP 服务器自我标识为 Librarian——这是你的代理与该角色通信的对象，这也是为什么暴露的工具被命名为 Librarian_SearchLibrary、Librarian_AddToLibrary 等。

概述

代理库为人工智能代理提供了文本、文档和知识的持久存储。代理可以通过语义和关键词搜索存储信息，并在对话间保持上下文。

特色

• 为人工智能智能体提供持久性知识存储
• SQLite 存储，支持向量搜索 sqlite-vec
• 使用FTS5进行全文搜索，配合BM25排名
• 结合语义和关键词匹配的混合搜索
• 多样化结果的最大边际相关性（MMR）
• 可配置嵌入模型（本地或兼容OpenAI的API）

Agent Library

A personal knowledge library for AI agents, built on Arcade for the Model Context Protocol (MCP).

Status. Agent Library is a community project built by Arcade.dev engineers. It is not an official Arcade.dev product and is not covered by Arcade.dev support or SLAs. See Support below for what to expect.

Naming. The package is Agent Library (agent-library on PyPI, librarian as the Python module). The MCP server inside it identifies itself as Librarian — that’s the role your agent talks to, and it’s why the exposed tools are named Librarian_SearchLibrary, Librarian_AddToLibrary, and so on.

Overview

Agent Library provides AI agents with persistent storage for text, documents, and knowledge. Agents can store information and retrieve it later through semantic and keyword search, maintaining context across conversations.

graph LR
    A[Agent Stores Info] --> B[Parser]
    B --> C[Chunker]
    C --> D[Embedder]
    D --> E[(SQLite + vec)]
    F[Agent Queries] --> G[Hybrid Search]
    E --> G
    G --> H[Relevant Context]

Features

• Persistent knowledge storage for AI agents
• SQLite storage with sqlite-vec for vector search
• Full-text search using FTS5 with BM25 ranking
• Hybrid search combining semantic and keyword matching
• Max Marginal Relevance (MMR) for diverse results
• Configurable embedding models (local or OpenAI-compatible API)
• Header-aware text chunking with overlap
• Time-bounded search filters
• CLI and MCP server interfaces

Multi-Modal Support

Agent Library supports indexing and searching across multiple file types:

Installation

The recommended way to install Agent Library is as a uv tool, which gives you the libr CLI and the MCP server in an isolated environment:

uv tool install 'agent-library[all]'

The [all] extra pulls in optional support for PDFs, images, OCR, and code-aware embeddings. If you’d rather install only what you need:

uv tool install agent-library                # core only (text + code)
uv tool install 'agent-library[pdf]'         # add PDF support
uv tool install 'agent-library[vision]'      # add image support
uv tool install 'agent-library[ocr]'         # add OCR for image-based PDFs

Installing from source

If you want to contribute or run from a clone:

git clone https://github.com/arcadeai-labs/agent-library.git
cd agent-library
./setup.sh