# Robo Voice Kill — Kill All Robo Voice Infrastructure

Kill all Robo Voice processes and free all ports in one shot.

Use when the user says "kill servers", "kill robo", "stop everything", "shut it down", or wants to clean up all Robo Voice infrastructure.

## What it does

Run these steps in sequence:

### Step 1: Kill all known process families

```bash
# Agent and its multiprocessing children
pkill -9 -f "uv run python src/agent.py" 2>/dev/null
pkill -9 -f "multiprocessing.resource_tracker" 2>/dev/null
pkill -9 -f "multiprocessing.spawn" 2>/dev/null
pkill -9 -f "multiprocessing.forkserver" 2>/dev/null

# Frontend
pkill -9 -f "next dev" 2>/dev/null
pkill -9 -f "pnpm dev" 2>/dev/null
pkill -9 -f "next-server" 2>/dev/null

# MCP server
pkill -9 -f "bun server" 2>/dev/null
```

### Step 2: Force-free all ports

```bash
lsof -ti:7890 | xargs kill -9 2>/dev/null  # MCP server
lsof -ti:3000 | xargs kill -9 2>/dev/null  # Frontend
lsof -ti:3001 | xargs kill -9 2>/dev/null  # Frontend (overflow)
lsof -ti:3002 | xargs kill -9 2>/dev/null  # Frontend (overflow)
```

### Step 3: Verify and report

Wait 1 second, then check everything is dead:

```bash
pgrep -af "agent.py" | grep -v pgrep
pgrep -af "next" | grep -v pgrep
lsof -ti:7890,3000,3001,3002
```

Print a clean report:

```
Agent:    dead
Frontend: dead
Port 7890: free
Port 3000: free
```

If anything survives, show the PID and kill it individually.

## Rules

- Always use kill -9 (SIGKILL) — these processes don't respond to graceful shutdown reliably
- Always verify after killing — don't trust pkill alone, the multiprocessing children often survive
- Report the final state clearly — the user needs to know ports are free before launching new sessions