Skip to content

ajbrown/007

Folders and files

NameName
Last commit message
Last commit date

Latest commit

 

History

10 Commits
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 

Repository files navigation

Project 007 - Elite Multi-Agent Development Plugin

     ___     ___    _____
    / _ \   / _ \  |___  |
   | | | | | | | |    / /
   | | | | | | | |   / /
   | |_| | | |_| |  / /
    \___/   \___/  /_/

   ACCELERATE YOUR VELOCITY

Transform your solo development workflow into an elite multi-agent operation. Let AI agents handle the repetitive work while you focus on architecture and innovation.

💡 The Problem

You're a skilled engineer, but you're constantly context-switching between:

  • Writing requirements → Planning implementation → Coding → Testing → Documenting
  • Bouncing between 5 open files trying to keep patterns consistent
  • Writing the same boilerplate for the 47th time this month
  • Waiting for CI to tell you about that test you forgot to update

Sound familiar? You need force multipliers, not more coffee.

✨ The Solution

Project 007 gives you a team of specialized AI agents that work autonomously:

  • Strategic Planner - Transforms requirements into detailed, executable implementation plans
  • Autonomous Executor - Deploys 3+ agents simultaneously to knock out tasks in parallel
  • Quality Assurance - Runs comprehensive test suites and fixes failing tests automatically
  • Specialized Engineers - Frontend, security, testing specialists available on-demand

Real Example: What You Get

Before Project 007:

You: "I need to implement user authentication"
Claude: "Sure! Let me write that for you..."
[2 hours later, you have code but no tests, incomplete docs, and security gaps]

After Project 007:

You: /write-implementation-plan
# Creates: 15 discrete tasks with acceptance criteria, dependencies, specs

You: /next-priority docs/plans/main.md 3
# Spawns 3 agents:
#   Agent 1: Database schema + migrations
#   Agent 2: API endpoints + rate limiting
#   Agent 3: UI components + validation
# [All complete in ~45 minutes with tests, docs, security review]

You: *Goes to lunch feeling like a 10x engineer*

Productivity multiplier: 3-5x on well-defined tasks.


🚀 Quick Start (5 Minutes)

1. Install the Plugin

claude plugin install ajbrown/007

Verify installation:

claude plugin list
# You should see "project-007" in the output

2. Try Your First Command

Open Claude Code and run:

/write-implementation-plan

This creates an implementation plan from your requirements. (Don't have requirements yet? That's fine—it'll help you create them!)

3. Experience the Force Multiplier

/next-priority

Watch as 3 AI agents simultaneously execute tasks from your plan. Each agent:

  • Works in isolated git worktrees (no conflicts!)
  • Writes tests for their changes
  • Updates documentation
  • Merges back to main when complete

📦 What's Included

🎯 Commands (Your Mission Control)

Command What It Does When to Use
/write-implementation-plan Transforms requirements into detailed execution plans with phases, milestones, and parallelizable tasks Start of sprint, new feature, refactor planning
/next-priority Deploys 1-N AI agents to autonomously execute top-priority tasks in parallel After planning, during sprints, for autonomous execution
/testing/fix-tests Runs test suite, identifies failures, deploys agents to fix all failing tests After merges, during CI troubleshooting, pre-release

🎭 Agents (Your Specialist Team)

These specialized agents work autonomously or can be invoked on-demand:

Agent Specialization Example Use
product-manager Requirements, planning, feature prioritization Strategic planning, roadmap creation
frontend-engineer React, UI/UX, design systems, accessibility Building interfaces, component work
security-reviewer OWASP, vulnerability assessment, penetration testing Security audits, pre-deployment reviews
quality-engineer Test development, coverage analysis, test strategy Writing tests, improving test quality
code-reviewer Code quality, best practices, refactoring PR reviews, code quality checks
linear-scrum-master Linear integration, sprint planning, ticket management Syncing Linear with implementation plans

Note: Agents are deployed automatically by commands, but you can also invoke them directly for specific tasks.


📖 Core Workflow: Plan → Execute → Ship

The Autonomous Development Cycle

# 1. PLAN - Create your strategic implementation plan
/write-implementation-plan docs/reqs/main.md docs/plans/sprint-5.md

# 2. EXECUTE - Deploy autonomous agents to work the plan
/next-priority docs/plans/sprint-5.md 3

# 3. REVIEW - Check the completed milestone, provide feedback
# (Review the PRs, test the functionality, gather learnings)

# 4. ITERATE - Continue execution or update plan based on learnings
/next-priority docs/plans/sprint-5.md 3

# 5. VALIDATE - Ensure quality before shipping
/testing/fix-tests

# 6. SHIP - Deploy with confidence
git push origin main

Key Principles

  1. Plan in Milestones - Each milestone should be demonstrable and valuable
  2. Execute in Parallel - Let agents handle independent tasks simultaneously
  3. Review at Boundaries - Check work between milestones, not constantly
  4. Iterate Continuously - Update plans as you learn, repeat the cycle

💎 When to Use Project 007

✅ Perfect For

  • Feature Development - Turn requirements into working code fast
  • Technical Debt - Systematically eliminate debt with parallel execution
  • Testing Initiatives - Boost coverage with dedicated test-writing agents
  • Refactoring Projects - Break down large refactors into safe, discrete tasks
  • Documentation Sprints - Update docs across your codebase efficiently
  • Prototyping - Rapidly build MVPs with multi-agent parallelization

⚠️ Not Ideal For

  • Exploratory Coding - When you're not sure what to build yet
  • Learning New Tech - When you need to understand deeply, not just ship
  • Single-File Changes - Overhead isn't worth it for trivial edits
  • Highly Coupled Work - Tasks with complex interdependencies

🎓 Learning Resources

For Beginners

New to Claude Code?

  • Claude Code Documentation
  • Start with /write-implementation-plan to see how planning works
  • Use concurrency=1 for /next-priority until you understand the workflow

New to Agentic Workflows?

  • Think of agents as specialized team members who work autonomously
  • Each agent has specific skills and works on isolated tasks
  • You provide direction, agents handle execution and details

For Advanced Users


🔧 Configuration & Customization

Adjusting Concurrency

Start conservative (1-3 agents), increase as you get comfortable:

# Conservative - one task at a time
/next-priority docs/plans/main.md 1

# Balanced - most common usage
/next-priority docs/plans/main.md 3

# Aggressive - for highly parallelizable work
/next-priority docs/plans/main.md 5

Custom Paths

All commands support custom paths:

# Custom requirements and plan locations
/write-implementation-plan docs/reqs/v2.md docs/plans/v2-plan.md

# Different implementation plans for different initiatives
/next-priority docs/plans/frontend-work.md 3
/next-priority docs/plans/backend-work.md 2

Integration with Your Workflow

Project 007 works alongside your existing tools:

  • Linear - Use linear-scrum-master agent for ticket sync
  • Git - All work happens in proper branches with clean commits
  • CI/CD - Agents respect your test suite and linting rules
  • Code Review - Work is organized into reviewable PRs

🆘 Troubleshooting

"The agents aren't working as expected"

  • Check your implementation plan - Agents need clear acceptance criteria
  • Review dependencies - Ensure task dependencies are correctly marked
  • Start with concurrency=1 - Debug with sequential execution first
  • Check git status - Ensure you have a clean working directory

"I'm not seeing the plugin commands"

# Verify plugin is installed
claude plugin list

# Reinstall if needed
claude plugin uninstall project-007
claude plugin install ajbrown/007

"Tasks aren't completing properly"

  • Ensure acceptance criteria are specific and measurable
  • Check that required files/dependencies exist in your project
  • Review agent output for blockers or errors
  • Try smaller, more focused tasks

🤝 Contributing

Contributions welcome! See CONTRIBUTING.md for guidelines.

Ways to contribute:

  • Add new specialized agents
  • Improve existing commands
  • Share your workflows and patterns
  • Report issues or suggest enhancements

📄 License

MIT License - See LICENSE for details.


⚡ Quick Links


"The name's Code... Claude Code." 🍸

Built with ❤️ for developers who want to ship faster.

Plugin Version: 1.0.0 Last Updated: 2026-01-16

About

No description, website, or topics provided.

Resources

License

Stars

Watchers

Forks

Packages

No packages published

Contributors 3

  •  
  •  
  •