AI & agents

AI and agents

How AI assistants can read these docs, validate .well files and convert well reports, through the MCP server, llms.txt and the welldot-converter skill.

AI assistants are often the fastest way to turn a scanned drilling report into data, or to write code against a file format. They are also confident when they are wrong. These tools give an assistant the real specification and the real validator, so its answers and its .well files can be checked instead of trusted.

MCP server

Connect Claude, Cursor, VS Code or another MCP client to search these docs, read any page and validate .well files.

llms.txt and Markdown

Every page as plain Markdown, an index for assistants, and the whole site in one file.

Converting reports

The welldot-converter skill turns a well report into a .well file, and checks it with the validator.

Two common cases

A geologist converting reports. The assistant reads a PDF or a photo of a drilling report, follows the welldot-converter skill to extract every section, writes the file, and runs validate_well on the MCP server until it passes. The geologist then reviews the result against the report.

A developer writing code. The assistant looks up fields and rules with search_docs and get_page instead of guessing, and validates the test files it writes. Without MCP, give it /llms.txt or the Markdown of the pages you are working from.

What the tools guarantee

The same validator. validate_well runs parseWell from @welldot/core, exactly as the welldot.org editor and the libraries do. It adds no rules of its own (Validating).
The real spec. search_docs and get_page read the same pages you are reading now, including the full specification.
Nothing stored. The MCP server is stateless and read-only. It does not keep the files you validate.

A valid file can still be wrong. Validation checks the shape of the data, not whether the numbers match the report. Always review a converted file against its source (Converting a report).

Converting reports with an assistant