THE BREAKOUT PROJECT

ClaudeGlobalCommands

CCGlobalCommands

A comprehensive collection of optimized AI agents and workflows for Claude Code. Prompt engineering as a product.

0 GitHub Stars
0 Forks
45% Token Savings
Python 89% PowerShell 9% Shell 2% Created Jul 3, 2025 164 KB MIT License
Tech Stack
Built With
Python
89%
PowerShell
9%
Shell
2%
Claude Code CLI
Playwright
Prompt Engineering
Architecture
Hierarchical Agent Orchestration
A multi-layered system where specialized agents are coordinated by orchestrators, achieving 45% token cost reduction through intelligent prompt routing.
Entry Points
/guide
/agents
/workflows
/execute
Orchestration
prompt-engineer
senior-engineer
documentation
code-review
Agents (30+)
css-safety-check
visual-testing
documentation-update
start-workflow
+26 more
Execution
Playwright Runner
Token Optimizer
Multi-Agent Coordinator
workflow-orchestration.py
# Hierarchical agent coordination pattern class WorkflowOrchestrator: def execute(self, workflow_name, context): # Route to specialized agent based on task agent = self.select_agent(workflow_name) # Token-optimized prompt injection prompt = self.compress_context(context) # 45% reduction # Multi-agent coordination results = agent.run(prompt, callbacks=[ self.css_safety_check, self.visual_regression_test, ]) return self.aggregate_results(results)
Commands
Slash Command Interface
Intuitive command-line interface with 7 primary commands that unlock 30+ specialized agents.
/guide
Interactive walkthrough for new users. Explains the system, suggests starting workflows, and adapts to skill level.
/agents
Browse and inspect all 30+ specialized agents. View capabilities, token costs, and usage examples.
/prompt-engineer
Meta-agent for crafting and optimizing prompts. Analyzes token usage and suggests compression strategies.
/senior-engineer
Senior-level code review agent. Checks architecture, patterns, edge cases, and suggests improvements.
/documentation
Auto-generates and maintains documentation. Syncs with code changes and validates examples.
/workflows
Browse 15+ multi-agent workflows. Chain agents together for complex tasks like full code reviews.
/execute
Run any workflow or agent directly. Supports parameters, dry-run mode, and verbose output.
Workflows
Multi-Agent Pipelines
Complex tasks decomposed into agent chains that run in sequence, with each step building on the last.
1 css-safety-check
→
2 visual-testing
→
3 code-review
→
4 documentation-update
→
✓ Complete
terminal
# Run the visual testing workflow with Playwright $ claude /execute visual-testing --target ./components [Agent: css-safety-check] Scanning 47 CSS files... [Agent: css-safety-check] All clear - no regressions detected [Agent: visual-testing] Launching Playwright... [Agent: visual-testing] Capturing 12 viewport snapshots [Agent: visual-testing] 12/12 passing [Agent: code-review] Analyzing changes... [Agent: code-review] 2 suggestions, 0 issues # Token usage: 3,420 (45% below baseline)
Key Features
What Makes It Special
⚡
30+ Specialized Agents
Each agent is purpose-built for a specific task -- from CSS safety checks to documentation generation. No bloat, no overlap.
💰
45% Token Reduction
Intelligent context compression and prompt routing dramatically cut API costs without sacrificing output quality.
🎨
Visual Testing
Playwright-powered visual regression testing ensures CSS changes never break the UI. Screenshot diffing across viewports.
🔗
Multi-Agent Workflows
15+ pre-built pipelines chain agents together. One command triggers a full review-test-document cycle.
📚
Self-Documenting
The documentation agent keeps README and guides in sync with code. Examples are validated automatically.
🛠
Hierarchical Orchestration
Orchestrator pattern routes tasks to the right agent automatically. Agents can delegate to sub-agents for complex work.
By The Numbers
Project Stats
0
GitHub Stars
0
Forks
0
Specialized Agents
0
Multi-Agent Workflows
0
% Token Savings
0
KB Codebase
The Story
What This Project Proved
"This repo proved that well-crafted prompts could be a product."
Prompt engineering at scale -- Writing one good prompt is easy. Building a system of 30+ prompts that work together, route intelligently, and stay maintainable is an entirely different discipline.
Token optimization matters -- A 45% reduction in token usage translates directly to cost savings. The compression techniques developed here became reusable patterns across other projects.
Multi-agent coordination patterns -- The hierarchical orchestration model (orchestrators routing to specialized agents) became the mental model for all subsequent agent work. Agents that delegate beat monolithic prompts every time.