Data, datasets and agents
The site's data in forms a program can read: open datasets, a read-only MCP server for AI assistants, Markdown versions of the pages, and an API for writing posts.
- Datasets
- /datasets — CSV and JSON, CC BY 4.0
- MCP server
- api.sweetvaluelab.com/mcp — read-only, no account
- Markdown
- Add .md to a page's address
- Account needed
- Only to write posts through the API
Open datasets
/datasets publishes the rows behind the stock index, the reverse DCF, the sector screeners, the sentiment gauge, the earnings calendar and the investor pages as CSV and JSON, rebuilt with the site every night. Each JSON file carries its date, its source, its licence and a description of every column. They are free to use under CC BY 4.0: credit SweetValueLab and link back.
The MCP server, for AI assistants
A read-only MCP server at https://api.sweetvaluelab.com/mcp answers the questions the pages answer, from the same data, with no account and no key. It has five tools:
| Tool | Answers |
|---|---|
search_stocks | Find a covered stock by name or ticker |
get_stock_valuation | A stock's price against its headline model, the other models, and its two grades |
get_market_context | The Treasury yield curve, the risk-free rate and equity risk premium the models discount with, and the sentiment gauge |
get_investor_holdings | A manager's latest 13F holdings and moves |
get_13f_consensus | The stocks the managers hold in common |
Each answer ends with the page it came from. The grades come back as the pages show them — a letter, the model and the date. Setup instructions for the common assistants are on the datasets page. In a browser that supports it, the site also offers the same tools to an assistant running in the page.
Markdown versions of the pages
Most pages an assistant is likely to read — the stock pages, the investor pages, the screeners, the manual and the blog — have a Markdown version at the same address with .md added. A request that asks for text/markdown gets it at the page's own address. /llms.txt describes the whole site for an assistant that has just arrived.
Embedding a stock
A stock's price chart has an Embed button: a small card with the close, the model value and the price against it, for another site to place in its own page.
The API
The public read endpoints the pages use are listed in /openapi.json. Writing a post for the blog through the API needs an account and a personal token; Write for us has the steps. Nothing else on the site needs an account.