The reference toolchain for the MTHDS open standard — editing, formatting, linting, and language-server support for .mthds and .toml files. Built on a Taplo fork. Ships a VS Code / Cursor extension, the plxt CLI, and the pipelex-tools PyPI package.
What is MTHDS? — An open standard for defining AI methods as typed, composable, human-readable files. A
.mthdsfile describes what an AI should do — its inputs, outputs, logic, and data types — in plain TOML that both people and machines can read. Pipelex is the runtime that executes them. Learn more at docs.pipelex.com.
The Pipelex CLI for formatting and linting MTHDS and TOML files. Install via PyPI:
pip install pipelex-tools
# or
uv add pipelex-tools| Command | Description |
|---|---|
plxt fmt |
Format TOML and MTHDS documents |
plxt lint |
Lint TOML and MTHDS documents |
plxt lsp stdio |
Start the language server (stdio transport) |
plxt get |
Extract a value from a TOML document |
plxt config |
Print default config or its JSON schema |
plxt completions |
Generate shell completions |
Configuration: plxt looks for .pipelex/plxt.toml (preferred) or plxt.toml in your project root (falls back to .taplo.toml).
First-class editing support for .mthds files and TOML — syntax highlighting, semantic tokens, formatting, completions, schema validation, and more.
code --install-extension Pipelex.pipelex
# or
cursor --install-extension Pipelex.pipelexSee editors/vscode/README.md for full details.
- MTHDS language support: Rich syntax highlighting, semantic tokens, and language features for
.mthdsfiles - Concept definitions:
[concept.Name]sections with specialized highlighting - Pipe definitions:
[pipe.name]sections for method steps - Jinja2 template syntax: Colorized highlighting for Jinja2 expressions in prompt fields
- Template variables:
@variableand$variablesyntax for inserting data into Jinja2 templates - Pipelex config discovery:
.pipelex/plxt.tomlorplxt.toml - Complete TOML 1.0.0 support and tooling
File all issues at this repository.
domain = "hr_screening"
description = "Analyze a job offer to build a scorecard, batch process CVs"
main_pipe = "screen_candidates"
[concept.Scorecard]
description = "Evaluation scorecard built from a job offer"
[concept.Scorecard.structure]
job_title = { type = "text", required = true }
company = { type = "text" }
required_skills = { type = "list", item_type = "text" }
criteria = { type = "list", item_type = "concept", item_concept_ref = "hr_screening.Criterion" }
[pipe.screen_candidates]
type = "PipeSequence"
inputs = { job_offer = "Document", cvs = "Document[]" }
output = "CvResult[]"
steps = [
{ pipe = "extract_job_offer", result = "job_pages" },
{ pipe = "build_scorecard", result = "scorecard" },
{ pipe = "evaluate_cv", batch_over = "cvs", result = "results" },
]
[pipe.build_scorecard]
type = "PipeLLM"
inputs = { job_pages = "Page[]" }
output = "Scorecard"
model = "claude-4.6-opus"
prompt = """Analyze this job offer and build a scorecard..."""See the MTHDS language reference for the full standard.
"Pipelex" is a trademark of Evotis S.A.S.
This is the repository for Taplo, a TOML v1.0.0 toolkit, more details on the website.
The project is very young, so bugs and incomplete features are expected, so any help is welcome!
The correctness of the TOML parsing and decoding is not yet entirely guaranteed (as there is no official 1.0.0 compliance test suite yet).
All kinds of contributions are welcome. Make sure to read the CONTRIBUTING.md first!