Coding standards, frameworks, refactoring, and code review (16.7K skills)
Apply broad software craftsmanship principles to architecture, coding, and team practices. The skill helps reduce duplicated knowledge and coupling, build a thin end-to-end slice, make assumptions explicit, and reason about estimates and reversible technology choices.
Harden a working but untested prototype through a staged process: add a safety net of tests, improve readability and structure, clarify business-rule boundaries, and address production reliability, scale, and data correctness. Decisions and progress are recorded for later sessions.
Make a working codebase faster and easier to change by measuring performance and protecting existing behavior with tests. The guided process addresses architecture, code structure, bottlenecks, data access, and reliability, recording baselines and decisions in project documents.
Diagnose slow web pages by examining network latency, resource loading, caching, browser rendering, and real-time connections. Use it to improve Core Web Vitals, reduce blocking resources, choose suitable HTTP and streaming approaches, and get specific fixes for failed diagnostic checks.
Use this skill to shape software around how a business actually works. It helps define bounded contexts, shared domain language, entities, value objects, aggregates, and domain events, making system boundaries and integrations easier to reason about.
Guide a new app from idea validation through product design and engineering decisions to a focused first release. The workflow asks for choices at each phase and records plans and findings in project documents for later sessions.
Use this skill to reason through database and data-system choices, from data models and storage engines to replication, partitioning, transactions, and batch or stream processing. It helps make trade-offs around scale, consistency, and reliability explicit.
Improve existing code structure while preserving its behavior. The skill maps code smells, such as long methods or tangled conditionals, to named refactorings and guides small, test-backed changes for safer maintenance and feature work.
Structure software so business rules stay independent of frameworks, databases, and user interfaces. The skill helps place code in layers, define boundaries, apply dependency inversion and SOLID principles, and review whether components can change independently.
Keep a Codex delegation to an installed implementer skill waiting in the background until its relay completes. It avoids model-driven progress polling, then returns the result for separate review. First use requires setting up its MCP server.
Delegate a bounded coding task to Muse Code, then inspect its diff and rerun project checks before committing. It supports sequential task queues and session-based follow-up work, with the orchestrator responsible for review and final commits.
Delegate a bounded coding task to Oh My Pi (`omp`), then inspect its changes and rerun the project’s checks before committing. It helps you manage implementation work or a queue of tasks while keeping review and final decisions in your hands.
Delegate a scoped coding task to the Cursor Agent CLI, then inspect the diff and rerun the project’s checks before committing. It supports implementation tasks and queues while leaving review and final integration with the orchestrator; read-only dispatch can also provide a second opinion without edits.
Delegate a bounded coding task to the OpenCode CLI, then inspect its changes and rerun the project’s checks before committing. It supports single tasks and sequential queues while keeping review and commit decisions with the orchestrating agent.
Delegate a coding task to the GitHub Copilot CLI, then review its changes and rerun project checks before committing. The workflow supports optional model and reasoning settings, read-only planning, and sequential task queues while keeping final review with the orchestrator.
Use this skill to delegate coding tasks or a task queue to the Google Antigravity CLI (`agy`). It guides the agent in writing a clear brief, reviewing returned changes, rerunning project checks, and committing verified work.
Assign a bounded coding task to the Cline CLI, inspect the resulting diff, and rerun project checks before committing. It provides a brief, background relay, and review workflow, including a planning mode for read-only work and support for sequential task queues.
Use this skill to delegate coding tasks to the Mistral Vibe CLI (`vibe`). It guides brief writing, change review, and independent project checks, with options for planning, controlled access, and follow-up work before committing.
Use this skill to delegate coding work to the Z.AI ZCode CLI. It helps prepare a focused brief, inspect the resulting diff, rerun project checks, and commit verified changes, with options for read-only reviews and follow-up work.
Send a bounded coding task to Warp’s headless Agent CLI, `oz`, then review the resulting diff and rerun project checks before committing. It can also manage sequential task queues while leaving implementation review and commit decisions to the orchestrator.
Delegate a bounded coding task or task queue to the Qoder CLI (`qodercli`) as a background implementer. The skill helps prepare a brief, run the relay, and review the changes and project checks before the orchestrator commits them.
Delegate coding work to Aider (`aider`) as a background implementer, including tasks using a local or self-hosted OpenAI-compatible model. The skill guides task briefing and independent diff review before the orchestrator commits changes.
Use this skill to delegate a coding task or task queue to the Command Code CLI (`cmd`). It helps prepare a focused brief, inspect the returned changes, rerun project checks, and commit verified work yourself.
Delegate a coding task or queue to the Kimi Code CLI (`kimi`) as a background implementer. The skill guides briefing, dispatch, and independent review of the diff and project checks before the orchestrator commits the work.