Standard Operating Procedures

This section contains step-by-step guides for common tasks and processes.

Core Processes

Client Management

• Client Onboarding - New client setup process
• Team Assignment - Assigning clients to teams
• Client Offboarding - Client departure process

Development

• TDD Pipeline - Test-Driven Development pipeline with specification agents
• Code Review - Code review standards and process
• Deployment - Deployment procedures
• Testing - Testing requirements and procedures

Operations

• Incident Response - How to handle incidents
• HALT Procedures - When and how to use HALT
• Monitoring - What to monitor and when to alert

Communication

• Notifications - Notification system usage
• Client Communication - Standards for client communication
• Escalation - When and how to escalate issues

HALT System

The HALT system is GE's safety mechanism for stopping work when needed.

When to HALT

• Security incident detected
• Data breach suspected
• Compliance violation identified
• Critical infrastructure failure
• Cross-team boundary violation

How to HALT

# System-wide HALT
touch /ge-ops/system/halt/HALT

# Agent-specific HALT
touch /ge-ops/system/halt/HALT-{agent_name}

# JSON HALT with details
cat > /ge-ops/system/halt/halt-$(date +%Y%m%d-%H%M%S).json <<EOF
{
  "status": "active",
  "scope": "system",
  "reason": "Security incident detected",
  "timestamp": "$(date -Iseconds)",
  "issued_by": "ron"
}
EOF

Clearing a HALT

Only Dirk-Jan or Ron can clear a HALT:

# Remove the HALT file
rm /ge-ops/system/halt/HALT*

# Or update JSON status
# Edit the JSON file and change "status" to "resolved"

Team Boundaries

CRITICAL: Once a client is assigned to a team, they stay with that team FOREVER.

• Team Alpha serves Team Alpha clients
• Team Beta serves Team Beta clients
• Cross-team work triggers automatic HALT
• Only Dolly can assign new clients to teams

This ensures context retention and prevents knowledge fragmentation.