Pipelex Blog — Executable AI methods, MTHDS language for Spec-Driven Development
Insights on repeatable AI workflows and the future of executable methods.
From Spec-Driven Development to Compilable Specs
MTHDS methods are already executable specs, interpreted by a runtime today. But a declarative language this high-level could also be compiled, deterministically generating code, tests, and guidance for agents. That would give us a compiler for specs, an engineering layer for Spec-Driven Development.
Why Multi-Step AI Workflows Need a New Language
MTHDS is an open, typed DSL for defining multi-step AI workflows. Methods are executable: install one, run it from your terminal, pipe it into another. Think of it as turning LLM workflows into proper UNIX tools. Pipelex is the MIT-licensed runtime, shipped as a Claude Code plugin.
The Know-How Graph
Skills gave agents reusable guidance: portable, natural-language instructions that steer behavior. But they have no types, no validation, no guaranteed output structure. That’s why we built MTHDS: an open standard for typed, validated AI workflows with deterministic orchestration.