Users need guidance on invoking external AI coding tools. This skill provides documented patterns and workflows for using Claude Code Task agents to call Codex, Gemini, and Cursor with proper authentication and read-only controls.

Managing multiple AI coding providers requires understanding their unique capabilities and command patterns. This skill provides documentation and workflows for delegating tasks to Claude, Codex, Gemini, and Cursor with proper authentication, read-only safeguards, and fallback strategies.

Manually tracking and updating package manifests wastes time and causes errors. This skill automatically detects new imports in your code and updates the correct manifest files with the right package names and version constraints.

Understanding unfamiliar codebases wastes hours of developer time. This skill provides structured patterns to quickly identify entry points, trace dependencies, detect architectural patterns, and catalog technical debt.

Working with LLM outputs often leads to parsing errors and type mismatches. This skill provides universal BAML patterns for type-safe LLM prompting with automatic code generation for Python and TypeScript.

Building MCP servers from scratch is complex without guidance on protocol requirements, tool design patterns, and evaluation strategies. This skill provides a comprehensive workflow for creating production-ready MCP servers in Python or TypeScript, covering research, implementation, review, and evaluation phases.

Building rich HTML artifacts for Claude conversations requires setup, dependencies, and bundling. This skill automates the entire workflow from project initialization to single-file bundle generation.

Creating custom slash commands requires understanding file format, YAML configuration, and dynamic features. This skill provides comprehensive guidance on building reusable, powerful commands with arguments, file references, and bash execution for Claude Code.

Shell scripts often fail due to unquoted variables, missing error handling, and unsafe patterns. This skill provides production-ready templates, security-aware best practices, and modern command-line alternatives to help you write maintainable Bash and Zsh scripts.

Complex scientific Python dependencies slow down research. Pixi unifies conda and PyPI ecosystems with fast dependency resolution, reproducible lockfiles, and seamless environment management for scientific computing projects.

Software breaks when requirements change because tests are missing or inadequate. This skill provides a disciplined red-green-refactor methodology that ensures every behavior change has verified test coverage before code is written.

Stop guessing at fixes and start finding real root causes. This skill provides a systematic debugging methodology that prevents wasted time on symptom fixes and repeated failed attempts.

Complex multi-task implementations often suffer from scope creep and quality issues. This skill provides a structured workflow with fresh subagents per task and dual-review gates (spec compliance then code quality) with diff-based validation to ensure every task is correctly implemented.

Code reviews catch bugs and quality issues before they reach production. This skill provides structured prompts and checklists for AI-powered code review of pull requests and merge requests.

Code review feedback often contains suggestions that may be unclear, technically questionable, or scope-expanding. This skill provides a systematic approach to evaluate feedback before implementing, ask clarifying questions when needed, and push back with technical reasoning when suggestions are wrong.

Development plans often fail when executed without structure or verification. This skill provides a disciplined methodology for implementing plans with batch processing and review checkpoints.

Fixing multiple unrelated test failures wastes time when done sequentially. This skill provides a proven workflow to identify independent problems, dispatch focused agents in parallel, and integrate fixes efficiently without conflicts.

Agents produce inconsistent output that fails downstream validation. This skill provides a standardized pattern for validating structured JSON against defined schemas before writing final output, ensuring data quality.

Running multiple AI tasks sequentially wastes time. This pattern lets you spawn parallel subagents to work concurrently, dramatically reducing total execution time for independent tasks.

Catch issues early with an iterative testing workflow. Run tests, lint, and typecheck after each code change before declaring work complete.