TinyDB MCP Server
Provides long-term memory for chatbots through a TinyDB-backed MCP server. Enables storing, searching, and managing JSON records with schema validation without requiring external databases.
README
tinydb-emcipi

This repo is docker-friendly – an MCP wrapper around TinyDB that gives chatbots long-term memory without leaving your laptop.
Tiny MCP server backed by TinyDB — drop-in long-term memory for chatbots and small-context LLM apps.
Why?
Keeping structured facts in the prompt is wasteful. With tinydb-emcipi an agent can store, search and mutate JSON records through MCP tool calls, loading only what it needs.
- Zero dependencies – TinyDB stores JSON on disk.
- Schema-first – every record validated; the schema itself is queryable.
- Opinionated tools – list DBs, switch DB, introspect schema, CRUD, search.
- Docker-friendly – run locally or in CI with a single
docker compose up.
Quick Start
Native
pip install -r requirements.txt
python -m main # MCP endpoint on http://localhost:8000/mcp
Docker Compose
docker compose up # builds dependencies at first run
The compose file mounts the repo into /app inside a python:3.12-slim container and starts the same command.
Available MCP Tools (summary)
| Category | Tool | Purpose |
|---|---|---|
| Discovery | create_database_tool |
create new DB (optionally with initial schema) |
list_databases_tool |
see all TinyDB files | |
switch_active_db_tool |
set the DB every other tool will use | |
get_database_info_tool |
schema, sample records, field list | |
| Records | search_records_tool |
field-value query, returns schema too |
get_all_records_tool |
dump every record | |
add_record_tool / update_record_tool |
create / modify validated records | |
| Schema | set_schema_tool |
create / update JSON Schema for a DB |
Project Structure (top-level)
config/ configuration & sample
logs/ runtime logs
sandbox/db/ TinyDB JSON databases
utils/ code modules (db, schema, mcp wrappers)
main.py entry-point
docker-compose.yml
requirements.txt
Contributing
Pull requests are welcome! Please:
- Follow PEP 8 and keep imports sorted.
- Add or update unit tests.
- Keep the README and doc-strings concise and useful.
Example MCP Prompt
A ready-to-use JSON request lives at scratchpad/mcp.json.sample showing how to invoke these tools from an MCP client.
© 2025 • MIT License
推荐服务器
Baidu Map
百度地图核心API现已全面兼容MCP协议,是国内首家兼容MCP协议的地图服务商。
Playwright MCP Server
一个模型上下文协议服务器,它使大型语言模型能够通过结构化的可访问性快照与网页进行交互,而无需视觉模型或屏幕截图。
Magic Component Platform (MCP)
一个由人工智能驱动的工具,可以从自然语言描述生成现代化的用户界面组件,并与流行的集成开发环境(IDE)集成,从而简化用户界面开发流程。
Audiense Insights MCP Server
通过模型上下文协议启用与 Audiense Insights 账户的交互,从而促进营销洞察和受众数据的提取和分析,包括人口统计信息、行为和影响者互动。
VeyraX
一个单一的 MCP 工具,连接你所有喜爱的工具:Gmail、日历以及其他 40 多个工具。
graphlit-mcp-server
模型上下文协议 (MCP) 服务器实现了 MCP 客户端与 Graphlit 服务之间的集成。 除了网络爬取之外,还可以将任何内容(从 Slack 到 Gmail 再到播客订阅源)导入到 Graphlit 项目中,然后从 MCP 客户端检索相关内容。
Kagi MCP Server
一个 MCP 服务器,集成了 Kagi 搜索功能和 Claude AI,使 Claude 能够在回答需要最新信息的问题时执行实时网络搜索。
e2b-mcp-server
使用 MCP 通过 e2b 运行代码。
Neon MCP Server
用于与 Neon 管理 API 和数据库交互的 MCP 服务器
Exa MCP Server
模型上下文协议(MCP)服务器允许像 Claude 这样的 AI 助手使用 Exa AI 搜索 API 进行网络搜索。这种设置允许 AI 模型以安全和受控的方式获取实时的网络信息。