How supersaiyane/loopengineering works
Loop Engineering is a Kiro IDE orchestration system that turns a prompt into appropriately sized engineering work: the hook classifies the task and selects an execution mode, loads steering and role-specific skills, then agents execute under quality gates. The documented workflow also covers discovery for ambiguous requests and persistence of lessons and decisions across sessions. The README describes the high-level wiring; source excerpts here are from a separate nested HealthGuard Pro project, so implementation-level links for Loop Engineering are limited to documented relationships.
Main entry points: loop-engine.md Plan Do Verify · elite-team-orchestrator.json Orchestrator Hook · skills Agent Roles · memory Session Memory
Components 17
Task Routing
- Orchestrator Hook
kiro/hooks/elite-team-orchestrator.json - Discovery Engine
kiro/steering/discovery-engine.md - Execution Modes
kiro/steering/orchestrator-routing.md
Agent Guidance
- Steering Rules
kiro/steering - Skill Selection
kiro/steering/skill-selection-matrix.md - Role Skills
kiro/skills
Execution and Quality
- Agent Roles
kiro/skills - Plan Do Verify
kiro/steering/loop-engine.md - Quality Gates
kiro/steering/session-guards.md - Risk Guards
kiro/steering/session-guards.md - Reuse Ladder
kiro/steering/ponytail.md
Persistent Context
- Session Memory
kiro/memory - Lessons
kiro/memory/lessons.md - Checkpoint
kiro/memory/checkpoint.md - Decisions
kiro/memory/decisions.md
Other components
- Engineer
- Engineering Work
Connections 16
- Engineer → Orchestrator Hook: submits prompt
README.md - Orchestrator Hook → Execution Modes: selects mode
README.md - Orchestrator Hook → Discovery Engine: invokes when unclear
README.md - Orchestrator Hook → Steering Rules: loads
README.md - Orchestrator Hook → Agent Roles: activates
README.md - Skill Selection → Role Skills: maps labels to
README.md - Agent Roles → Role Skills: uses
README.md - Agent Roles → Plan Do Verify: executes
README.md - Plan Do Verify → Quality Gates: runs checks
README.md - Risk Guards → Plan Do Verify: requires confirmation
README.md - Reuse Ladder → Plan Do Verify: guides reuse
README.md - Plan Do Verify → Engineering Work: produces
README.md - Plan Do Verify → Session Memory: updates
README.md - Session Memory → Lessons: records lessons
README.md - Session Memory → Checkpoint: records progress
README.md - Steering Rules → Decisions: preserves decisions