Skip to content

Skills Overview

Skills are reusable workflows that provide structured approaches to common development tasks. They encode best practices and ensure consistent, high-quality work.

How to Use Skills

In Claude Code

Skills are invoked automatically when relevant, or explicitly:

Use the debugging skill to investigate this issue

In Other Platforms

See Platform Support for platform-specific invocation methods.

Skill Categories

Core Workflow Skills

Foundational skills for structured development (from obra/superpowers):

Skill When to Use
brainstorming Before coding - explore requirements and design
writing-plans After brainstorming - create implementation plan
executing-plans Execute a written plan systematically
test-driven-development Implementing any feature or fix
debugging Unified debugging entry point - routes to appropriate methodology
using-git-worktrees Isolating feature work from main codebase
finishing-a-development-branch Complete development work with merge/PR/cleanup options

Code Quality Skills

Skills for maintaining and improving code quality:

Skill When to Use
auditing-green-mirage Auditing test suite quality
fixing-tests Fixing failing or weak tests
fact-checking Verifying claims and assumptions
finding-dead-code Identifying unused code
requesting-code-review Requesting structured code review

Feature Development Skills

Skills for building and reviewing features:

Skill When to Use
develop End-to-end feature implementation
reviewing-design-docs Reviewing design documents
reviewing-impl-plans Reviewing implementation plans
devils-advocate Challenging assumptions and decisions
merging-worktrees Merging parallel worktrees
resolving-merge-conflicts Resolving git merge conflicts with synthesis

Specialized Skills

Domain-specific skills:

Skill When to Use
async-await-patterns Writing async JavaScript/TypeScript

Meta Skills

Skills about skills and subagent orchestration:

Skill When to Use
using-skills Understanding how to invoke and use skills
writing-skills Creating new skills
writing-commands Creating new commands
instruction-engineering Effective prompt engineering for subagents and LLMs
optimizing-instructions Reducing token usage in instruction files
documenting-tools Writing documentation for MCP tools and APIs
dispatching-parallel-agents Parallel subagent orchestration
smart-reading Reading files/output without blind truncation

Creating Custom Skills

See Writing Skills for instructions on creating your own skills.

Personal skills placed in ~/.claude/skills/ take priority over spellbook skills.