Agent Deploy

This slash command provides a templated set of instructions asking an AI agent to generate an agent-specific version of a deployment script found in a code repository.

Deployment scripts often generate verbose outputs.

This verbosity poses a challenge to context load when ingested by an agent or a sub-agent who mop up lots of unnecessary lines of Bash output just to ultimately know a simple success or failure message.

The objective of creating agent-specific deployment scripts is to create an alternative deployment pathway specifically intended for use by AI agents, bypassing this otherwise useful verbosity.

The human can execute the human version, or the agent can when there is something to debug, but for routine deployments, the streamlined, verbosity-minimizing version can be used instead.

The slash command provided instructs the agent to create a refactored version of the script for this purpose and to update its own internal references to refer to it. An example is provided of a before and after deployment script.

The Solution

Two deployment scripts:

deploy.sh - Verbose (Human-Friendly)

Full output for manual debugging and monitoring. Shows every step of:

• Git commit/push
• Docker build progress
• Image transfer
• Migration execution
• Container startup

agent-deploy.sh - Minimal (AI-Friendly)

Outputs only key milestones:

Deployment started
Committing to GitHub...
Building image...
Transferring to VM...
Restarting containers...
Deployment complete
Status: OK - http://YOUR_VM_IP:3000

Full logs are written to /tmp/deploy.log for debugging if needed.

Usage

# Full deploy
./agent-deploy.sh        # Minimal output (for AI sessions)
./deploy.sh              # Verbose output (for humans)

# Quick restart (container only)
./agent-deploy.sh quick
./deploy.sh quick

Configuration

Edit the configuration section at the top of each script, or set environment variables:

export VM_HOST="192.168.1.100"
export VM_PORT="3000"
export REMOTE_DIR="/home/deploy/app"
export IMAGE_NAME="myapp:latest"
export DB_CONTAINER="app-db"
export DB_USER="postgres"
export DB_NAME="appdb"

Features

• Git integration: Auto-commits and pushes before deploy
• Docker build: Builds image locally with build args (commit hash, build time)
• Image transfer: Saves and loads Docker image to remote VM via SSH
• Config sync: Transfers docker-compose.yml and .env files
• Migrations: Detects, transfers, and runs SQL migrations automatically
• Migration archival: Moves applied migrations to applied/ folder and commits
• Health check: Verifies deployment success

Requirements

• Docker (local machine)
• SSH access to remote VM (key-based auth recommended)
• Docker and docker-compose on remote VM
• Git repository

Directory Structure

your-project/
├── Dockerfile
├── docker-compose.yml
├── .env.production
├── deploy.sh
├── agent-deploy.sh
└── migrations/
    ├── 001_initial.sql
    ├── 002_add_users.sql
    └── applied/           # Auto-created, holds completed migrations

License

MIT