spec-kitty-mission-system

Designing AI agent workflows with state machine guard conditions involves mapping mission types to required artifacts and advancement rules. You can model mission.yaml state machines to define step ordering, required artifacts, and doctrine-driven behaviors for reliable AI execution runs.

Built-in mission types like software-dev, research, plan, and documentation define end-to-end step sequences, transitions, and required artifacts. They structure workflow behavior by mapping specific work domains to runtime DAGs and isolated execution workspaces.

Selecting the right mission type requires mapping your goal to the correct workflow domain. You evaluate built-in mission types, their specific step sequences, and required artifacts to determine which aligns with your work domain before resolving templates through the five-tier chain.

Doctrine composition works by delegating step contracts to procedures, tactics, and directives. Action indices scope injected doctrine per action, ensuring the correct workflow behaviors are applied and advanced through the mission hierarchy from work packages to isolated execution workspaces.

The five-tier chain for resolving mission templates is the mechanism used to map mission phases to required artifacts and guards. It ensures workflow templates are correctly resolved and applied across different work domains, runtime DAGs, and execution contexts.

You should avoid a mission system for workflow design when your task does not require strict step ordering, required artifacts, or doctrine-driven behaviors. It is best suited for complex AI execution runs needing reliable end-to-end workflow behavior and advancement rules.