What problem does it solve? Agents working inside an AI-driven development loop often know what step comes next but not why the loop is shaped that way, so they act incorrectly in cases the state table does not enumerate. This Skill is the universal preload that carries the loop's intentional design — its intake chain, state machine, iteration axis, and inner-loop rules — to every dispatched persona. ## Core Features & Use Cases - State machine reference: Canonical transition table covering filed, description closed, ready, in progress, drafted, reviewed, and blocked states across product, content, and loop lanes, naming who acts and what artifact records each transition. - Loop model and mode selection: Guidance for choosing between gitflow-multi-env and trunk-single-env promotion models, and reading the scrum versus kanban mode from the record rather than inferring it from query output. - Design intent and discipline framing: Explains Agent Harness Engineering — mechanical gates over memory, the three surfaces (cadence, gates, harness-as-artifact), and when to dispatch agents-lead for machinery changes. - Use Case: When picking up a slice, proposing a change to the loop itself, or writing publicly about Agent Harness Engineering, load this Skill so decisions align with the loop's documented intent rather than ad-hoc judgment. ## Quick Start Ask the agent to apply the agents-configuration skill to explain which persona closes the description and applies the ready label for a loop-typed issue before dispatching any work.
