Commands¶

CLI commands and shell usage reference.

Core Commands¶

coding¶

Launch a coding agent with all integrations (Docker services, LSL, health monitoring, constraints).

# Launch specific agents
coding --claude      # Claude Code (Anthropic OAuth / Max subscription)
coding --opencode    # OpenCode (auto-selects provider based on network)
coding --copilot     # GitHub Copilot CLI
coding --copi        # Alias for --copilot

# Specify project directory
coding --project ~/my-project

# Clean start (kills all coding processes, frees all ports, then launches)
coding --force
coding --force --claude

# Dry run (validates config, network, Docker — does not launch)
coding --opencode --dry-run

# Help
coding --help

Network-Aware Agent Selection¶

The launcher automatically detects your network environment and configures each agent accordingly:

Network-Aware Agent Selection

Agent	Inside VPN (proxy)	Outside VPN (direct)	Provider
`--claude`	Works via proxy	Works direct	Anthropic (OAuth / Max)
`--opencode`	GitHub Copilot Enterprise	Anthropic direct	Auto-selected
`--copilot`	Works via proxy	Works direct	GitHub Copilot CLI

Auto-configuration:

Inside VPN: proxy auto-detected (proxydetox on 127.0.0.1:3128), all APIs routed through proxy
Outside VPN: proxy env vars cleared, direct connections used
OpenCode switches model: github-copilot-enterprise/claude-opus-4.6 (VPN) → claude-opus-4-6 (public)
Each agent validates its API endpoint is reachable before launch

`--force` Flag¶

Performs a full cleanup before startup — use when ports are stuck or orphaned processes prevent launch:

Stops all Docker coding containers (docker compose down)
Kills process supervisors and health monitors (prevents respawning)
Kills all processes on coding ports (3030-3033, 3847-3850, 8080, 9090, 12435)
Kills remaining coding-repo node processes
Verifies all ports are free
Proceeds with normal startup

# When "address already in use" errors block startup
coding --force

# Equivalent to a clean reboot of all coding services
coding --force --claude

vkb¶

View Knowledge Base visualization.

# Open in browser
vkb

# Version info
vkb --version

# Debug mode
vkb --debug

Testing Commands¶

Installation Test¶

# Check-only mode (default, safe)
./scripts/test-coding.sh

# Interactive repair
./scripts/test-coding.sh --interactive

# Auto-repair coding-internal issues
./scripts/test-coding.sh --auto-repair

# Help
./scripts/test-coding.sh --help

LSL Validation¶

# Validate LSL configuration
node scripts/validate-lsl-config.js

# Generate repair script
node scripts/validate-lsl-config.js --generate-repair

Docker Commands¶

Start/Stop Services¶

# Start all services
docker compose -f docker/docker-compose.yml up -d

# Stop all services
docker compose -f docker/docker-compose.yml down

# Restart
docker compose -f docker/docker-compose.yml restart

View Logs¶

# All services
docker compose -f docker/docker-compose.yml logs -f

# Specific service
docker compose -f docker/docker-compose.yml logs -f coding-services

Rebuild¶

# Rebuild after code changes
docker compose -f docker/docker-compose.yml build --no-cache
docker compose -f docker/docker-compose.yml up -d

# Remove volumes (data loss!)
docker compose -f docker/docker-compose.yml down -v

Health Checks¶

# All MCP health endpoints
for port in 3848 3849 3850; do
  echo "Port $port: $(curl -s http://localhost:$port/health | jq -r '.status')"
done

# LSL monitor health (Phase 33+: read from coordinator, not .health/*.json)
curl -fs http://localhost:3034/health/state \
  | jq '.lsl | to_entries | map(select(.key | endswith(":coding"))) | .[0].value | {status, lastBeat}'

# Container status
docker compose -f docker/docker-compose.yml ps

Knowledge Base¶

Within Claude Code Session¶

# Incremental analysis
"ukb" or "update knowledge base"

# Full analysis
"full ukb" or "fully update knowledge base"

Command Line¶

# Purge entities from date
node scripts/purge-knowledge-entities.js 2025-12-23

# Dry run
node scripts/purge-knowledge-entities.js 2025-12-23 --dry-run

# With options
node scripts/purge-knowledge-entities.js 2025-12-23 --team=coding --verbose

LSL Recovery¶

# Recover LSL files from transcripts
PROJECT_PATH=/path/to/project CODING_REPO=/path/to/coding \
  node scripts/batch-lsl-processor.js from-transcripts ~/.claude/projects/-path-to-project

# Recover specific date range
PROJECT_PATH=/path/to/project CODING_REPO=/path/to/coding \
  node scripts/batch-lsl-processor.js retroactive 2024-12-01 2024-12-03

Git Submodules¶

# Update all submodules
git submodule update --remote

# Initialize missing submodules
git submodule update --init --recursive

# Update specific submodule
git submodule update --remote integrations/mcp-server-semantic-analysis

Debugging¶

Enable Debug Output¶

# LSL debugging
DEBUG_STATUS=1 TRANSCRIPT_DEBUG=true node scripts/enhanced-transcript-monitor.js

# Full debug mode
DEBUG_LSL=1 node scripts/generate-proper-lsl-from-transcripts.js --mode=foreign --verbose

Process Management¶

# Clean start - kill ALL coding processes and restart fresh
coding --force

# Find running monitors
ps aux | grep enhanced-transcript-monitor

# Kill monitors
pkill -f enhanced-transcript-monitor

# Restart monitoring
coding --restart-monitor