mthds-ai

Build new AI method from scratch using the MTHDS standard (.mthds bundle files). Use when user says "create a pipeline", "build a workflow", "new .mthds file", "make a method", "design a pipe", or wants to create any new method from scratch. Guides the user through a 10-phase construction process.

Check and validate MTHDS bundles for issues. Use when user says "validate this", "check my workflow", "check my method", "does this .mthds make sense?", "review this pipeline", "any issues?", "is this correct?". Reports problems without modifying files. Read-only analysis.

Edit existing MTHDS bundles (.mthds files). Use when user says "change this pipe", "update the prompt", "rename this concept", "add a step", "remove this pipe", "modify the workflow", "modify the method", "refactor this pipeline", or wants any modification to an existing .mthds file. Supports automatic mode for clear changes and interactive mode for complex modifications.

Explain and document MTHDS bundles. Use when user says "what does this pipeline do?", "explain this workflow", "explain this method", "walk me through this .mthds file", "describe the flow", "document this pipeline", "how does this work?", or wants to understand an existing MTHDS method bundle.

Fix issues in MTHDS bundles. Use when user says "fix this workflow", "fix this method", "repair validation errors", "the pipeline is broken", "fix the .mthds file", after /mthds-check found issues, or when validation reports errors. Automatically applies fixes and re-validates in a loop.

Prepare inputs for MTHDS methods. Use when user says "prepare inputs", "create inputs", "use my files", "generate test data", "template", "synthesize inputs", "mock inputs", "I have a PDF/image/document to use", "make sample data", or wants to create inputs.json for running a .mthds pipeline. Handles user-provided files, synthetic data generation, placeholder templates, and mixed approaches. Defaults to automatic mode.

Build a method bundle top-down by stepwise refinement — capture the whole job as one pipe signature, then refine it layer by layer into a runnable method that is valid at every step.

Prepare inputs for MTHDS methods. Use when user says "prepare inputs", "create inputs", "use my files", "generate test data", "template", "synthesize inputs", "mock inputs", "I have a PDF/image/document to use", "make sample data", or wants to create inputs.json for running a .mthds pipeline. Handles user-provided files, synthetic data generation, placeholder templates, and mixed approaches. Defaults to automatic mode.

Build new AI method from scratch using the MTHDS standard (.mthds bundle files). Use when user says "create a pipeline", "build a workflow", "new .mthds file", "make a method", "design a pipe", or wants to create any new method from scratch. Guides the user through a 10-phase construction process.

Check and validate MTHDS bundles for issues. Use when user says "validate this", "check my workflow", "check my method", "does this .mthds make sense?", "review this pipeline", "any issues?", "is this correct?". Reports problems without modifying files. Read-only analysis.

Edit existing MTHDS bundles (.mthds files). Use when user says "change this pipe", "update the prompt", "rename this concept", "add a step", "remove this pipe", "modify the workflow", "modify the method", "refactor this pipeline", or wants any modification to an existing .mthds file. Supports automatic mode for clear changes and interactive mode for complex modifications.

Explain and document MTHDS bundles. Use when user says "what does this pipeline do?", "explain this workflow", "explain this method", "walk me through this .mthds file", "describe the flow", "document this pipeline", "how does this work?", or wants to understand an existing MTHDS method bundle.

Fix issues in MTHDS bundles. Use when user says "fix this workflow", "fix this method", "repair validation errors", "the pipeline is broken", "fix the .mthds file", after /mthds-check found issues, or when validation reports errors. Automatically applies fixes and re-validates in a loop.

Prepare inputs for MTHDS methods. Use when user says "prepare inputs", "create inputs", "use my files", "generate test data", "template", "synthesize inputs", "mock inputs", "I have a PDF/image/document to use", "make sample data", or wants to create inputs.json for running a .mthds pipeline. Handles user-provided files, synthetic data generation, placeholder templates, and mixed approaches. Defaults to automatic mode.

Set up Pipelex inference configuration — choose backends and configure API keys. Use when user says "set up pipelex", "configure backends", "configure inference", "set up API keys", "pipelex setup", "pipelex init", wants to run methods for the first time, or gets a config/credential error when running.

Manage MTHDS packages — initialize, configure exports, list, and validate. Use when user says "init package", "set up METHODS.toml", "manage packages", "mthds init", "validate package", "list package", or wants to manage MTHDS package manifests.

Bump the minimum required versions that mthds-agent enforces at runtime — the Claude Code mthds plugin, pipelex (and pipelex-agent, same package), plxt, and the Codex app. Use when the user says "bump required versions", "bump min pipelex version", "bump minimum plxt version", "raise minimum plugin version", "bump min codex version", "update required versions", "set min pipelex to X.Y.Z", "bump min mthds plugin version", or any variation of changing one or more of these floors. Trigger even when the user names only one of the targets — this skill handles partial updates.

Show the minimum required versions that this mthds-agent release enforces at runtime — the Claude Code mthds plugin, pipelex (and pipelex-agent, same package), plxt, and the Codex app. Read-only: reports the current floors without changing anything. Use when the user says "check min versions", "what are the required versions", "show minimum versions", "what's the min pipelex version", "what's the min codex version", "current version floors", "which plxt version do we require", "show required versions", or any variation of asking what the current minimums are. This is the read-only counterpart to the bump-required-versions skill — if the user wants to *change* a floor, use that one instead.

Detect interface-contract drift between this package's code and a baseline (defaults to the last release tag, but the user can specify any tag or commit). Compares the source files behind the mthds-js provided/consumed/implemented interfaces — the mthds-agent CLI, the MTHDS Protocol API runner, plxt/pipelex-agent passthrough, and the hook pipeline — against the interface specs in ../docs/specs/. Use when the user says "check the contract", "contract review", "contract check", "did we break the contract", "check interfaces", "API contract", "protocol drift", "compare to vX.Y.Z", or before shipping/releasing a version that touches CLI, API-runner, protocol, or passthrough code. Also trigger automatically before the /release skill runs, and proactively when you notice such changes during a PR review.

Prepare a new release for the mthds-js project. Bumps version in package.json, syncs package-lock.json, updates CHANGELOG.md, manages the release/vX.Y.Z branch, runs checks, and commits. Use when the user says "release", "prepare a release", "bump version", "new version", or "cut a release".

Prepare a new release for the MTHDS project. Bumps version in pyproject.toml, syncs uv.lock, updates CHANGELOG.md, manages the release/vX.Y.Z branch, validates docs build, and commits. Use when the user says "release", "prepare a release", "bump version", "new version", or "cut a release".

Automates the mthds-python release workflow: bumps the version in pyproject.toml, finalizes the CHANGELOG.md Unreleased section, runs quality checks, regenerates uv.lock, creates a release/vX.Y.Z branch, commits, pushes, and opens a PR to main. Use when user says "release", "cut a release", "bump version", "prepare a release", "make a release", "ship it", "create release branch", or any variation of shipping a new version of the mthds Python package. The user can optionally provide changelog content inline when invoking the skill (e.g. "/release Added a new base runner option"), which will be used as the changelog entry for this version.

Scaffold a new method for `mthds-std` end-to-end: design → bundle → package → benchmark selection → dataset loader → scorers → eval wiring → docs → `make cc`. Does NOT run the evaluation (that costs LLM credits and is the user's call). Use when the user says: "add a method to mthds-std", "new std primitive", "create a method in mthds-std", "scaffold <verb> method", "add <foo> to the standard library", or asks to start a new primitive with its eval setup. Do NOT use this skill for: - methods that belong in `pipelex/methods` (vertical solutions) — that's a different repo - modifying an existing method (use `/mthds-edit` instead) - running existing experiments (use `make run-experiment` directly)