PAVED docs for the AI agent era

Documentation 📚

Welcome to the pave docs. You can read them, but they’re not really for you - they’re for the human+agent pair doing the work. 🤝

Overview

pave is a CLI tool for creating, validating, and managing PAVED documentation - a framework optimized for AI agent workflows.

Command Description
pave init Initialize pave in your project
pave new <type> <name> Scaffold a new document
pave prompt <type> Generate AI agent prompts
pave index Generate documentation index
pave check Validate documents against rules
pave config Manage configuration
pave adopt Scan existing docs to help onboard

Guides 📖

Guide Description
Getting Started Set up pave in a new project
Onboarding Existing Projects Adopt pave in an existing codebase
CI/CD Integration Run pave in CI pipelines

Document Types 📑

pave supports three core document types:

Components 🔧

For services, libraries, and modules. Includes Purpose, Interface, Configuration, Verification, Examples, Gotchas, and Decisions sections.

pave new component auth-service

Runbooks 📋

For operational procedures. Includes When to Use, Preconditions, Steps, Rollback, Verification, and Escalation sections.

pave new runbook deploy-production

ADRs 📝

Architecture Decision Records. Includes Status, Context, Decision, Consequences, and Alternatives Considered sections.

pave new adr use-rust

Philosophy 🧠

“Docs aren’t for humans anymore; they’re for a human+agent pair doing work.”

The PAVED framework treats documentation like APIs:

  • Precise contracts - not vague prose
  • Small surfaces - one concept per doc
  • Versioned - track changes
  • Validated - enforce quality rules
  • Optimized for retrieval + execution - agents need ground truth

Learn more in the Manifesto.