
AKShare
io.github.hangeaiagentPublished by GitHub account hangeaiagentv0.1.1Updated Oct 24, 2026
Let a model explore AKShare's 1000+ China market data functions on its own — search, inspect, call
Overview
A local MCP server that lets an assistant search, inspect and call AKShare's 1000+ China market data functions on demand.
- What you can do
- Search AKShare's function library by keyword to find which functions can fetch a given dataset, such as 龙虎榜 or 北向资金 holdings details. Inspect a function's signature to learn the parameters it expects. Call a chosen function with arguments and get its returned data, with column names left in their original Chinese. Responses are capped by a configurable row limit, and truncated results say so explicitly.
- When to use it
- Useful when you need China market, macro, futures, fund or company-governance data from AKShare and do not want to enumerate every interface in advance. Best run on a machine inside mainland China, since most upstream interfaces are there.
- Requirements
- Runs as a local process, installed from the PyPI package with uvx, or built and run as a Docker image (about 1.5G, several minutes to build). Optional environment variables AKSHARE_MCP_HOST, AKSHARE_MCP_PORT and AKSHARE_MAX_ROWS (default 200). Network access to upstream data sites, preferably from within China or via a proxy. No authentication or API key.
Installation
In SourceWeft
- Open AKShare in the dashboard and add it to a workspace.
- Enable the server for the chats that should use its tools.
Desktop only via STDIO. STDIO servers start a local process, so they need the SourceWeft desktop host.
Other MCP clients
Follow the launch instructions in the repository.
README
AKShare MCP
把 AKShare 变成一个你自己跑的能力,接进 Hunter。
为什么不是「数据源」
数据源的模型是一条记录 = 一个 (市场, 数据类型, 接口地址)。
AKShare 有一千多个函数 —— 龙虎榜、十大股东、券商研报、公司治理、 南向资金、宏观、期货、基金……按数据源建模就得枚举一千多条,做不到。
Hunter 原来只从里面挑了 6 条进目录,而那 6 条是我们用得上的,不是你 用得上的。你想查「北向资金持股明细」而我们没挑,你就没辙。
改成能力之后,模型拿到三个工具,自己去找、自己去调:
不需要我们提前枚举。
为什么要你自己跑
AKShare 是 Python 库不是 HTTP 服务,没法在表单里填一个 URL。而且它多数 接口的上游在国内,从境外容器直连经常打不通。
跑起来
镜像大约 1.5G(AKShare 自己拖着 pandas/lxml/beautifulsoup4 一大串), 首次构建要几分钟。
验证:
接进 Hunter
能力 → 接入一个工具 → 填:
Hunter 跑在 Docker 里的话,localhost 指的是 Hunter 自己的容器 ——
要填宿主机地址(Linux 上是 172.17.0.1,Docker Desktop 上是
host.docker.internal),或者把两个服务放进同一个 compose 网络。
⚠️ **不要填 Hunter 官方的任何地址。**这个服务存在的意义就是让你不依赖我们。
环境变量
AKSHARE_MAX_ROWS 不要调太大:AKShare 有些接口一次几万行,原样塞进模型
上下文会把真正有用的东西挤掉。被截断时返回里会明确说,不会假装那就是全部。
安全
akshare_call 只能调 akshare 模块里公开的、可调用的属性:
- 名字带
_前缀的拒绝 - 带
.的(想访问子模块)拒绝 - 不存在或不可调用的拒绝
- 不做
eval/exec/ 动态 import
它不是通用代码执行入口。
**但仍然不要暴露到公网。**这个服务没有鉴权,而 AKShare 会向第三方站点发 请求 —— 公网上任何人都能借你的机器发请求。绑内网,或者加一层带鉴权的反代。
常见问题
调用超时 / 连接被拒 多数 AKShare 接口的上游在国内。如果这台机器在境外,大部分接口都会失败 —— 把它跑在国内的机器上,或者给容器配代理。
函数名找不到
AKShare 的命名多是英文缩写,硬猜会一直失败:龙虎榜是 stock_lhb_detail_em
不是 stock_dragon_tiger。先 akshare_search 再 akshare_call。
返回的列名是中文 AKShare 的原样返回,我们不改列名 —— 改了之后你对着它的文档会对不上。
Source: tools/akshare-mcp/README.md at commit 4af6d96
Tools
0Version history
1- v0.1.1LatestSep 16, 2026
Score
50% of the applicable checks met
How complete this listing is, computed from the registry entry, the endpoint and public sources. Not a rating of the server and not a statement that it is safe or working today: reachability is a dated check, and no installs or reviews go into the score.
- Reached in the last 30 days25 pts
Keep the endpoint answering the MCP handshake; SourceWeft probes it every 30 days.
- Auth declared10 pts
Declare the credential headers in the registry entry, or keep the endpoint answering the probe so its auth can be classified.
- Tools listed15 pts
Answer tools/list without a credential, or ship a package that starts in a sandbox without one.
- DescriptionRequired15 pts
- Logo10 pts
- Documentation15 pts
- License10 pts