A curated collection of specialized skills for Claude Code, designed to enhance productivity, enforce best practices, and enable advanced workflows.
This repository contains custom Claude Code skills that extend Claude's capabilities with specialized knowledge, workflows, and tool integrations. Each skill provides deep expertise in a specific domain.
Systematic debugging workflow for issues spanning UI, backend, and database layers. Combines browser testing, log analysis, code examination, and automated server restarts to iteratively resolve full-stack issues.
Use cases:
- Debugging errors that span multiple application layers
- Resolving failing schedulers, import errors, database issues
- Systematic root cause analysis and verification
- API payload problems and configuration errors
- Issues originating in backend but manifesting in UI
Workflow:
- Detection - Multi-layer error detection (browser UI, backend logs, database state)
- Analysis - Root cause identification through code and log analysis
- Fix - One issue at a time with proven fix patterns
- Restart - Automated server restart with cache clearing
- Verification - Browser, API, database, and log verification
- Iteration - Continue until all issues resolved
MCP Integration: Integrates with robo-trader-dev MCP tools for 95%+ token-efficient debugging:
analyze_logs- Pattern detection with time windows (98% savings)check_system_health- Database, queues, API status (97% savings)diagnose_database_locks- Correlate logs with code patterns (95% savings)suggest_fix- Known pattern matching with examples (90% savings)
Triggers: Debugging full-stack issues, error states in UI, backend failures, database errors, cascading failures
Apply Edward Tufte's data visualization principles from "The Visual Display of Quantitative Information" to create high-impact presentations.
Use cases:
- Designing presentations with clarity and precision
- Creating charts and graphs that maximize data-ink ratio
- Reviewing slides for visual excellence
- Transforming complex data into clear visual displays
Key principles:
- Maximize data-ink ratio (remove chartjunk)
- Show the data with graphical integrity
- Use small multiples and sparklines for dense information
- Direct labeling instead of legends
- Eliminate decoration that doesn't inform
Triggers: "make a slide", "create presentation", "design chart", "visualize data", "review my slides"
A complete two-agent development system based on the research paper "Effective Harnesses for Long-Running Agents". Separates project planning from implementation for complex, multi-session development work.
Architecture:
- Opus (orchestrator) - Delegates work, never implements directly
- Initializer Agent (sonnet) - Breaks down projects into 200+ trackable features
- Coding Agent (sonnet) - Implements features one at a time with quality assurance
Features:
- Comprehensive project breakdown with feature tracking
- Enforcement hooks to prevent premature implementation
- Session state recovery after crashes
- Progress tracking across sessions
- Git integration for systematic commits
Use cases:
- Setting up new complex projects
- Managing multi-session development work
- Enforcing separation between planning and implementation
- Tracking progress on large feature sets
Triggers: "set up two-agent system", "install agent harness", "configure Opus delegation"
Each skill can be installed separately by copying its directory to your Claude Code skills location:
# For user-specific skills
cp -r <skill-name> ~/.claude/skills/
# Or for project-specific skills
cp -r <skill-name> /path/to/project/.claude/skills/The two-agent harness includes an automated setup script:
bash ./two-agent-harness/scripts/setup-two-agent-system.shThis will install agents, hooks, and reference documentation automatically.
Skills are automatically available in Claude Code once installed. Claude will invoke them when:
- You explicitly request the skill by name
- Your query matches the skill's trigger phrases
- The task context aligns with the skill's domain
Example:
User: "Create a presentation about Q4 sales data"
Claude: [Automatically invokes tufte-slide-design skill]
Each skill follows this structure:
skill-name/
├── SKILL.md # Skill definition and documentation
├── references/ # Supporting documentation and resources
├── assets/ # Templates, examples, or other assets
└── scripts/ # Automation scripts (if applicable)
To create a new skill:
- Create a directory with a descriptive name (kebab-case)
- Add a
SKILL.mdfile with frontmatter:--- name: skill-name description: Brief description of what the skill does and when to use it --- # Skill Name [Detailed documentation]
- Add reference materials in
references/directory - Include templates or examples in
assets/if needed
- Clear triggers: Define specific phrases that should invoke the skill
- Focused scope: Each skill should have a single, well-defined purpose
- Rich references: Include comprehensive documentation and examples
- Automated setup: Provide setup scripts for complex installations
Contributions are welcome! To contribute:
- Fork the repository
- Create a new skill or enhance an existing one
- Test the skill thoroughly with Claude Code
- Submit a pull request with clear documentation
Future skills under consideration:
- API design patterns
- Database schema design
- Security review workflows
- Code review checklist
- Testing strategy generator
These skills are provided as-is for use with Claude Code. Each skill may reference external resources that have their own licenses.
- Claude Code Documentation
- Creating Custom Skills Guide
- Effective Harnesses Research Paper
- Edward Tufte's Visual Display of Quantitative Information
For issues or questions:
- Open an issue in this repository
- Check individual skill documentation
- Refer to Claude Code's official documentation
Note: These skills require Claude Code to be installed and configured. They extend Claude's capabilities but do not modify the core Claude Code system.