moai-core-rules

Delegation-only execution separates command orchestration from task execution by routing all work through agents and skills. Alfred's 3-Layer Architecture enforces this pattern: commands coordinate, agents decide routing, and skills perform work, preventing commands from bypassing the architecture.

TRUST gates enforce five quality dimensions: tests, readability, unity, security, and traceability. These gates validate output before produce step in the 4-Step ADAP workflow, ensuring governance rules applied to decision-making and execution meet architectural standards.

ADAP—Analyze, Design, Assure, Produce—structures agent decision-making across command orchestration and skill execution. Each step enforces explicit multi-layer orchestration with stateless skills, ensuring consistent governance and traceability through the workflow.

Core constraints include delegation-only execution in commands, explicit multi-layer orchestration across all layers, stateless skill design, and mandatory TRUST quality gate validation before produce. These prevent architectural bypass and enforce consistent governance.

Tag-chain validation and commit standards apply throughout agent workflows to enforce traceability and consistency. They validate state transitions, decision chains, and skill outputs against governance rules before downstream propagation.

Yes. The 3-Layer Architecture and ADAP workflow support multi-agent orchestration and cross-skill coordination through explicit delegation patterns. Agent-First delegation ensures each agent operates within defined governance boundaries with consistent TRUST validation.