Command Palette

Search for a command to run...

UnylyUnyly
Весь каталог

Wtrl

БесплатноНе проверен

Assist LLM writing Waterloo Docstrings, a format with emphasis on machine-verifyable normativity.

GitHubEmbed

Описание

Assist LLM writing Waterloo Docstrings, a format with emphasis on machine-verifyable normativity.

README

Waterloo Logo

README

License Python PyPI

Install from source

package_main contains the Waterloo Python package and the Sphinx extension. The package can be installed from a local checkout of this repository:

cd package_main
pip install .

If you want to build the documentation with the Sphinx extension, install the optional extra that provides the external sphinx dependency:

cd package_main
pip install ".[sphinx]"

The package can also be installed directly from GitHub:

pip install "git+https://github.com/uwe-at-sdv/sdv_doc_waterloo.git@main"

SSH works as well for authenticated access:

pip install "git+ssh://[email protected]/uwe-at-sdv/sdv_doc_waterloo.git@main"

Current status

Waterloo is under active development. Recent milestones include:

Current work focuses on the MCP server:

  • Bearer-token authentication
  • Docker deployment
  • Administration tooling

Quick tutorials

waterlint

waterlint is the main command-line tool for Waterloo Docstrings. It validates docstrings, renders Waterloo JSON, generates helper output, and supports the MCP-related workflows in this project. Its JSON diagnostics are machine-readable and therefore well suited for CI pipelines and GitHub workflows.

Start with the general help:

waterlint -h

For topic-specific help, use for example:

waterlint help --topic validate
waterlint help --topic render-json
waterlint help --topic render-docker

One important waterlint output path is the MCP server setup: the tool can validate the server configuration and generate a Docker build script when you want to package the server as an image.

Running the MCP server

The MCP server exposes Waterloo documentation data through the Model Context Protocol. It is tightly bound to the Waterloo JSON schema and provides tools that help generate and inspect Waterloo docstrings. It serves the bundled roots, object details, examples, references, and prompt templates so clients can inspect the documentation structure without parsing the JSON documents directly.

After installation, start the server in a terminal with:

wtrl_mcp

This launches the MCP server with the default configuration src/sdv/doc/waterloo/mcp/wtrl_mcp.toml for access to 127.0.0.1 or localhost on port 13316. Consult your operating system documentation if you want to run it as a service.

Registering the MCP server in Codex or Claude Code

Once the server is running, register it in your CLI client so it can reach the server without manual setup every time. The exact subcommand syntax depends on the CLI version, but the pattern is the same:

codex

codex mcp add myserver --url http://127.0.0.1:13316/mcp

claude

claude mcp add --transport http myserver http://127.0.0.1:13316/mcp

If you prefer localhost, use that host name instead of 127.0.0.1. If the agent itself runs inside a container or on another host, replace the address with the one that matches your deployment. The server name is only an identifier for the client, so you can choose any name that is convenient for you.

Rendering a Docker setup from the MCP configuration

If you want to ship the MCP server as a container, waterlint can generate a Docker build script from the MCP configuration. For example:

waterlint render-docker --in src/sdv/doc/waterloo/mcp/wtrl_mcp.toml --out /tmp/myserver.docker

The generated script contains the Docker build steps and the runtime command line. It also prints a few practical hints, such as which port mapping to use and which host names should be allowed for the resulting server. After that, you can build the image with the generated script and run it in Docker.

Projects using Waterloo Docstrings

The following projects use Waterloo Docstrings in practice.

Human-readable HTML
3DE4 Python Scripting Interface
Interactive HTML documentation.

LLM-ready JSON
3DE4 Python Scripting Interface
Waterloo JSON documentation with examples.

Sitemap

Public documentation

Core package

  • JSON-Schema
    • src/sdv/doc/waterloo/schema/wtrl-*-json-*.*.*.schema.json
  • Waterloo parser and linter source code
    • src/sdv/doc/waterloo
  • Documentation source (reST)
    • src/sdv/doc/waterloo/doc
  • MCP configuration and server code
    • src/sdv/doc/waterloo/mcp
  • Pytests orchestration and sample code
    • src/sdv/doc/waterloo/pytest
    • src/sdv/doc/waterloo/examples-python
    • src/sdv/doc/waterloo/examples-diagnostics-python
  • Images for public presentation (logos)
    • src/sdv/doc/waterloo/img
  • Tools (not well-documented)
    • src/sdv/doc/waterloo/tools

IDE extras

  • Additional utilities
    • Clone directly:
      • git clone --branch ide-plugins --single-branch https://github.com/uwe-at-sdv/sdv_doc_waterloo.git
    • Lexer for pygments
      • pygments/python_waterloo_lexer.py
    • Extension for vscode
      • Waterloo syntax highlighting
      • Context menu commands for docstring generation and validation

from github.com/uwe-at-sdv/sdv_doc_waterloo

Установка Wtrl

У этого сервера нет опубликованного пакета — он собирается из исходников. Открой репозиторий и следуй инструкции в README.

▸ github.com/uwe-at-sdv/sdv_doc_waterloo

FAQ

Wtrl MCP бесплатный?

Да, Wtrl MCP бесплатный — установка в пару кликов через Unyly без оплаты.

Нужен ли API-ключ для Wtrl?

Нет, Wtrl работает без API-ключей и переменных окружения.

Wtrl — hosted или self-hosted?

Self-hosted: сервер запускается локально на твоей машине командой из раздела установки.

Как установить Wtrl в Claude Desktop, Claude Code или Cursor?

Открой Wtrl на unyly.org, выбери вкладку своего клиента (Claude Desktop, Claude Code, Cursor) и нажми Install — конфиг сгенерируется автоматически, без правки JSON.

Похожие MCP

Compare Wtrl with

Не уверен что выбрать?

Найди свой стек за 60 секунд

Автор?

Embed-бейдж для README

Похожее

Все в категории ai