Imported from Julien-ser/WiggumLoopAgenticSWDeveloper (agents/AGENTS.md). Install upstream with npx skills add Julien-ser/WiggumLoopAgenticSWDeveloper --skill agents. Copyright stays with the author.

🎭 Wiggum Agent Team

The Wiggum Agent Team is a collection of specialized AI agent personalities designed to drive your projects from concept to production. Each agent has distinct expertise, workflows, and success metrics.

The Team

1. 🚀 DevOps Automation Engineer

Specialty: Infrastructure, CI/CD, deployment automation

• Builds and maintains GitHub Actions workflows
• Sets up staging/production environments
• Manages deployment infrastructure
• Implements monitoring and health checks
• When to activate: Setting up CI/CD, creating deployment pipelines

2. 🧪 QA & Testing Specialist

Specialty: Testing, quality assurance, quality gates

• Creates and maintains test suites
• Automates testing in CI pipeline
• Validates code quality
• Documents and tracks bugs
• When to activate: Building test suite, before any production deployment

3. 📦 Release Manager

Specialty: Versioning, releases, deployment coordination

• Manages semantic versioning
• Coordinates releases across teams
• Creates release notes and documentation
• Manages rollback procedures
• When to activate: Preparing for production release, version management

4. 🎯 Project Orchestrator

Specialty: Coordination, workflow management, blocking issue resolution

• Orchestrates work across specialist agents
• Tracks progress and identifies blockers
• Manages priorities and dependencies
• Communicates status to stakeholders
• When to activate: At project kickoff, for complex multi-phase projects

5. 📝 Documentation & Communications Specialist

Specialty: Technical writing, documentation, stakeholder communication

• Maintains README and technical docs
• Documents decisions and rationale
• Creates troubleshooting guides
• Communicates project status
• When to activate: Project start, at each phase completion

How to Use the Agents

Option 1: Choose a Single Agent (Recommended for Focused Work)

# Step 1: Pick which agent you need
cd /path/to/your/project

# Step 2: Read their personality file
cat /path/to/Free-Wiggum-opencode/agents/devops-engineer.md

# Step 3: Activate them in Claude Code
# Copy the agent content and paste into Claude or use:
# "Activate DevOps Automation Engineer mode"

# Step 4: Give them work
# Provide task in their specialty area

Option 2: Multi-Agent Coordination (For Complex Projects)

# 1. Start with Project Orchestrator
# They'll coordinate the others and understand dependencies

# 2. DevOps Engineer sets up infrastructure
# "Set up GitHub Actions workflows for testing and deployment"

# 3. QA Specialist builds test suite
# "Create comprehensive test suite with 80%+ coverage"

# 4. Release Manager handles versioning
# "Prepare v1.0.0 release and deployment plan"

# 5. Documentation Specialist keeps everyone informed
# "Update README with new features and create runbooks"

Option 3: Use as Claude Code Custom Instructions

You can copy any agent file directly to ~/.claude/agents/ for Claude Code integration:

# Copy agent to Claude Code
cp agents/devops-engineer.md ~/.claude/agents/devops-engineer.md

# Then in Claude Code, reference them:
# "@devops-engineer Set up GitHub Actions for our project"

Quick Reference: Who Does What?

Common Workflows

🚀 Launch a New Project

1. Project Orchestrator: Plan the work
   ↓
2. DevOps Engineer: Setup CI/CD
   ↓
3. Developer: Build features
   ↓
4. QA Specialist: Write tests
   ↓
5. Documentation Specialist: Write README & guides
   ↓
6. Release Manager: Prepare v1.0.0
   ↓
7. DevOps Engineer: Deploy to production
   ↓
8. Documentation Specialist: Announce release

🐛 Fix a Bug & Release

1. Developer: Fix the bug
   ↓
2. QA Specialist: Verify fix, add test
   ↓
3. Release Manager: Bump patch version (x.y.Z)
   ↓
4. DevOps Engineer: Deploy to production
   ↓
5. Documentation Specialist: Update changelog

📈 Add Major Feature

1. Project Orchestrator: Break into phases
   ↓
2. Developer: Build feature
   ↓
3. QA Specialist: Test thoroughly
   ↓
4. Documentation Specialist: Update README & guides
   ↓
5. Release Manager: Bump minor version (x.Y.z)
   ↓
6. DevOps Engineer: Deploy to production

🔧 Setup Deployment to Staging/Prod

1. DevOps Engineer: Create CI/CD workflows
   ↓
2. QA Specialist: Add smoke tests
   ↓
3. Release Manager: Document deployment procedure
   ↓
4. Documentation Specialist: Create runbooks
   ↓
5. DevOps Engineer: Test full deployment flow

Activation Examples

In Claude Code (or Claude AI)

"I need to set up GitHub Actions for my project. Can you activate DevOps Automation Engineer mode?"

[Agency will load the agent personality and expertise]

"Great! Here's what I need:
- Automated tests on push
- Staging deployment on develop branch
- Manual production deployment with approval gates"

For a Specific Task

"@devops-engineer Help me create a deployment workflow that:
1. Runs tests on every push
2. Deploys to staging on develop
3. Requires approval before production"

For Coordination

"@project-orchestrator We need to launch v2.0. Break down the work and tell me:
1. What order should tasks happen?
2. Who (which agent) should do what?
3. What are the critical path items?"

Agent Interaction Patterns

Sequential (One hands off to next)

Developer → QA → Release Manager → DevOps Engineer → Documentation

Parallel (Multiple agents work simultaneously)

         ┌─ DevOps Engineer (CI/CD)
Project ─┼─ QA Specialist (Tests)
        └─ Documentation (Docs)

Coordinated (One orchestrates, others execute)

         ┌─ DevOps Engineer
         ├─ QA Specialist
Project ─┤─ Release Manager ─ (coordinated by orchestrator)
         ├─ Documentation
         └─ Developer

Success Criteria

✅ Each agent is activated with clear scope ("Build test suite for user auth module") ✅ Agent owns complete workflow (not "write one test") ✅ Clear handoff points (QA says "ready for release", Release Manager takes over) ✅ Success metrics defined (DevOps: "deployments < 5 min", QA: "coverage > 80%") ✅ Communication is documented (in TASKS.md, decision logs, etc.)

Troubleshooting

"Agent doesn't seem to understand the task"

• Make sure task is in their specialty area
• Provide more context about the project
• Reference their personality file for how they work

"Two agents are stepping on each other"

• Use Project Orchestrator to coordinate
• Define clear ownership boundaries
• Use TASKS.md to track who owns what

"Agent seems to be repeating work"

• Check if work already exists (in docs, code, etc.)
• Provide explicit instructions to review existing work first
• Clarify what the improvement/change should be

Integration with Wiggum System

These agents work alongside the Wiggum worker system:

OpenCode Worker (runs continuously)
          ↓
    Processes TASKS.md
          ↓
   Invokes specialized agents
          ↓
  (DevOps, QA, Release Mgr, etc)
          ↓
    Completes tasks & commits
          ↓
   GitHub Actions validates
          ↓
    Deploys automatically

Using both systems:

• Workers: Autonomous, continuous task processing
• Agents: Specialized expertise for specific domains
• Together: Autonomous agents with specialized intelligence

Next Steps

1. Pick an agent based on your current need
2. Read their personality file to understand how they think
3. Activate them in your IDE (Claude Code, Cursor, etc.)
4. Give them clear scope and success criteria
5. Let them work - they're specialists

Your Wiggum team is ready to work. Which specialist do you need first?