🏗️ Architecture

Orchestrates agent workflow for non-trivial tasks: plan-first mode, subagent use, self-improvement loops, verification before done, and autonomous bug fixing. Use for any task with 3+ steps, architectural decisions, bug reports, or when the user corrects the agent. Ensures plans go in tasks/todo.md, lessons in tasks/lessons.md, and changes stay minimal and provably correct.

Intelligent skill orchestrator that automatically finds, creates, executes, and improves skills. When you need to accomplish a task, this skill searches for existing skills (internal, GitHub via MCP, web), creates new skills if none found, executes them, and reviews execution to improve skills based on actual usage. Also handles feedback about skill-generated outputs - if you want to fix/adjust an output AND improve the skill that created it, invoke this with your feedback. Use when you want automated skill discovery, continuous improvement, or to provide feedback on previous skill outputs.

Orchestrate complex development with AUTOMATIC parallel subagent execution, continuous dispatch scheduling, dependency analysis, file conflict detection, and IOSM quality gates. Analyzes task dependencies, builds critical path, launches parallel background workers with lock management, monitors progress, auto-spawns from discoveries. Use for multi-file features, parallel implementation streams, automated task decomposition, brownfield refactoring, or when user mentions "parallel agents", "orchestrate", "swarm", "continuous dispatch", "automatic scheduling", "PRD", "quality gates", "decompose work", "Mixed/brownfield".

omni-frontend-expert is the ultimate frontend design skill. It's used to create distinctive, production-grade frontend interfaces with deeply considered aesthetics, flawless consistency, and professional interaction quality. Use when building any frontend; components, pages, apps, dashboards, tools. Enforces high standards across visual quality, UX behavior, design systems, accessibility, and performance while rejecting all forms of AI slop or lazy default design.

Run parallel code reviews with multiple AI agents, then synthesize into one report. Triggers on "review code" or "multi-agent review".

Design, build, and deploy Ask Gallery — a semantic photo search system for mobile devices. Use when asked to create image search features using natural language queries, build photo analysis pipelines with VLM/CLIP/OCR/face recognition, design vector search architectures with TiDB/Meilisearch/Kafka/EKS, or implement swappable ML model registries. Triggers on keywords like semantic search, photo search, image captioning, visual question answering, gallery AI, image embedding, multimodal search.

>

>-

>

Analyze Rails application upgrade path. Checks current version, finds latest release, fetches upgrade notes and diffs, then performs selective upgrade preserving local customizations.

Multi-agent autonomous startup system. Triggers on "Loki Mode". Takes PRD to deployed product with minimal human intervention. Requires --dangerously-skip-permissions flag.