Munk AI
Back to Docs
Concepts

Core Concepts

Understand the main workflows, runtime boundaries, and long-term assets behind Munk AI.

Core Concepts

Munk AI is not a single automation command. It is a local-first verification system built around three workflows:

  • start / run for execution on real targets
  • review / verify change for change-driven validation
  • serve / recording for local recording, inspection, and debugging

One runtime, multiple entry points

CLI, Local API, and MCP share the same orchestration core. This keeps behavior consistent across different entry layers.

Evidence over slogans

The system is designed to produce observations, findings, and artifacts that can be inspected later. The goal is not only to run actions, but to ground verification in evidence.

Long-term assets and run outputs

Some outputs are short-lived run artifacts. Others, like reusable test assets and recordings, can outlive a single run and become part of a more durable verification workflow.