CNEquity MCP
MCP lets an agent read a local Chinese-market archive and inspect financial data coverage, provenance and dates
- 958 ★ on the repository
- Last commit
- Apache-2.0
Extended access
Why this level
- Local archive query results enter the chosen model’s context
- Explicitly enabled live mode makes external requests that can trigger source restrictions
Install
- Works in
Manual install
pip install cnequityInstalls the package without a populated archive. Separately run `cne init --profile demo` for a small test collection.
Access and checks
How we reviewExtended accessThe tool runs code or makes network calls and only reads external data.
- Local archive query results enter the chosen model’s context
- Explicitly enabled live mode makes external requests that can trigger source restrictions
This is third-party code. Review the repository files before installing.
What it does
CNEquity connects an agent to a local Chinese stock-market archive. Separate commands collect prices and financial reports into Parquet, while `cne mcp` exposes them for reading. MCP tools cannot start collection or delete research data. Before querying, the agent can inspect available datasets and coverage dates. It then requests prices, fundamentals or SQL results. Responses report selection limits and source information so partial output is not mistaken for a complete dataset. Historical research needs an as-of date and a corporate-action adjustment policy. CNEquity returns those parameters with the result. The `live` mode can query a source without an existing archive, but only supports symbol lookup and unadjusted daily prices.
Who it is for. Chinese-equity analysts building a local archive of prices and financial reports for agent queries.
Good fit when
- You want to research collected prices without downloading them for every question
- You need to inspect coverage dates and completeness before calculation
- Fundamental comparisons need an explicit historical as-of date
- An agent needs a restricted SELECT query over a Parquet archive
Not a fit when
- You need a broker interface for placing orders
- You require exchange Level 2 data or complete trade-by-trade history
- You only need one current quote without maintaining an archive
Example request
Use CNEquity to inspect daily data available for the specified equity. Show dates and sources, then compare prices with an explicit adjustment policy.MCP
- Transport
- stdio, http
- Authentication
- not required
Limitations
No archive is bundled; collection needs Python 3.10 or later, networking and storage for history, staging and versions. Providers determine financial-data rights and can restrict commercial use or redistribution. Check migrations when updating the 0.x package because its format changes. Archive updates retain the latest five generations by default and can remove older ones; MCP does not invoke these operations, but separate collection commands do.
How to disable. Remove CNEquity from the client and stop any separately running HTTP process. Use `pip uninstall cnequity` to remove the Python package. Preserve configuration and needed data generations. Disable any separately installed daily collection schedule in your system scheduler.
FAQ
How do I install the package and create a small archive?
Run `pip install cnequity`, then `cne init --profile demo`. The second step queries real sources and saves five stocks over the latest 30 trading days in a separate directory.
Why must the configuration path be absolute?
The MCP client may launch the process from another directory. Pass the full path with `--config` so the agent uses the intended archive.
Can the SQL tool modify data?
It accepts one SELECT statement. The connection is restricted to data directories, with external extensions disabled. These DuckDB controls do not replace operating-system process isolation.
Are historical fundamentals always exact as of the chosen date?
No. `query_fundamentals` requires `as_of`, but default `best_effort` mode includes later backfilled rows. Check `pit_is_exact`; strict mode can return less data.
What if the required dataset has not been collected?
The server reports missing data rather than switching to `live` automatically. Check coverage with `describe_lake` before deciding on a separate collection.
Do sources require registration?
Not for basic collection; some optional sources require authorization. Code uses Apache-2.0, with separate data-use rights.
Discussion
No questions or reviews yet.
Related
MCP serversA local database of China A-share and ETF market data with an MCP server for natural-language queries
Financial Datasets MCP
Financial Datasets MCP Server
An MCP server for the Financial Datasets API: company financials, stock and crypto prices and news for the agent
The MCP server gives agents daily stock screens, relative-strength ratings, industry rankings and market breadth.
A hosted MCP server for financial data: quotes, filings, macro statistics and FTShare's own factors, read-only
A server feeds agents quotes for A-shares, Hong Kong and US stocks, indicators, funds, futures and options, plus four skills.
An akshare-based MCP server with quotes, indicators and news for China, Hong Kong and US stocks, plus cryptocurrency data
README in short
The archive supports MCP, Python and SQL access. Collection and maintenance use separate commands. Documentation distinguishes source provenance, retrieval dates and applied transformations.