Fly Deployment Manager

Comprehensive Fly.io deployment management through conversation.

Core Principle

Deployments should be simple and observable.

Guide users through Fly.io operations with clear feedback, helpful context, and actionable next steps. Abstract away complexity while maintaining full control.

Usage

/fly-deploy                      # Show deployment status
/fly-deploy status               # Detailed app and database status
/fly-deploy deploy               # Deploy the app
/fly-deploy logs                 # Stream recent logs
/fly-deploy secrets              # Manage environment secrets
/fly-deploy scale                # Scale machines and resources
/fly-deploy db                   # Database management
/fly-deploy ssh                  # SSH into a machine
/fly-deploy health               # Check health status
/fly-deploy restart              # Restart all machines
/fly-deploy destroy              # Destroy app (with confirmation)

Commands

/fly-deploy (default: status)

Shows current deployment status:

━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━
JFL PLATFORM - FLY.IO STATUS
━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━

App: jfl-platform
Region: sjc (San Jose)
URL: https://jfl-platform.fly.dev

MACHINES
  ✓ 78d40e1a - running (sjc)
  ✓ 90e3f7a4 - running (sjc)

DATABASE
  ✓ jfl-platform-db - running
  Postgres 15
  Primary: sjc

HEALTH
  ✓ All checks passing

━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━
QUICK ACTIONS
━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━
/fly-deploy deploy    Deploy latest changes
/fly-deploy logs      View recent logs
/fly-deploy ssh       SSH into machine

Implementation:

flyctl status --json
flyctl info --json
flyctl machines list --json

/fly-deploy deploy

Deploys the application:

Steps:

1. Check if there are uncommitted changes (warn if so)
2. Show what will be deployed (git commit hash/message)
3. Run deployment
4. Stream build/deploy logs
5. Show final status with URL
6. Suggest next action (view logs, test URL, etc.)

Implementation:

# Check for uncommitted changes
git status --porcelain

# Get current commit
git log -1 --oneline

# Deploy
flyctl deploy

# Check health after deploy
flyctl status --json

Example output:

Deploying jfl-platform...

Commit: a4a4dfd "auto: session save"
Branch: main

━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━

Building...
✓ Docker build complete (2m 15s)

Deploying...
✓ Machines updated (45s)

Health checks...
✓ All checks passing

━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━
DEPLOYED
━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━

https://jfl-platform.fly.dev

Want me to:
- Open the URL in your browser?
- Stream logs?
- Check health?

/fly-deploy logs [--follow] [--machine=ID]

Shows recent application logs:

Options:

• --follow / -f - Stream logs in real-time
• --machine=ID - Logs from specific machine
• --lines=N - Number of recent lines (default: 100)

Implementation:

# Recent logs
flyctl logs --lines 100

# Follow logs
flyctl logs --follow

# Specific machine
flyctl logs --machine 78d40e1a

Example output:

━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━
LOGS - jfl-platform
━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━

2026-01-19T18:45:23Z [info] ✓ Ready in 1204ms
2026-01-19T18:45:45Z [info] GET /api/auth/session 200 in 45ms
2026-01-19T18:46:12Z [info] POST /api/projects 201 in 123ms
2026-01-19T18:46:15Z [error] Database connection timeout

Last 4 lines. Use /fly-deploy logs --follow to stream.

/fly-deploy secrets

Manage environment secrets:

Subcommands:

• list - Show all secret names (not values)
• set <KEY>=<VALUE> - Set/update a secret
• unset <KEY> - Remove a secret
• sync - Sync from local .env file

Implementation:

# List secrets
flyctl secrets list

# Set secret
flyctl secrets set KEY="value"

# Unset secret
flyctl secrets unset KEY

# Bulk set from .env (with confirmation)
flyctl secrets import < .env

Example output:

━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━
SECRETS - jfl-platform
━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━

AUTH_SECRET                     Set 2d ago
DATABASE_URL                    Set 5d ago (from Fly Postgres)
GITHUB_CLIENT_ID                Set 5d ago
GITHUB_CLIENT_SECRET            Set 5d ago
NODE_ENV                        Set 5d ago

5 secrets total

━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━
ACTIONS
━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━
/fly-deploy secrets set KEY=value    Add/update secret
/fly-deploy secrets sync             Sync from .env

For sync command, confirm with user:

This will sync all secrets from your local .env file:

Will set/update:
  - AUTH_SECRET
  - GITHUB_CLIENT_ID
  - GITHUB_CLIENT_SECRET
  - STRIPE_SECRET_KEY

Continue? (yes/no)

/fly-deploy scale

Scale machines and resources:

Options:

• machines <count> - Scale to N machines
• memory <size> - Change memory (256mb, 512mb, 1gb, 2gb)
• cpu <count> - Change CPU count (1, 2, 4, 8)
• regions <list> - Add/remove regions

Implementation:

# Scale machines
flyctl scale count 3

# Scale memory
flyctl scale memory 1gb

# Scale CPU
flyctl scale cpu 2

# Scale to regions
flyctl regions add lax iad

Example output:

Current scale:
  Machines: 2
  Memory: 256mb
  CPU: 1 shared
  Regions: sjc

What would you like to change?
1. Scale to 3 machines (faster, costs ~$15/mo more)
2. Increase memory to 512mb (better performance)
3. Add regions (lax, iad - lower latency)
4. Custom configuration

/fly-deploy db

Database management:

Subcommands:

• status - Database status and connection info
• connect - Open psql connection
• migrate - Run Prisma migrations
• backup - Create manual backup
• restore <snapshot> - Restore from backup

Implementation:

# Database status
flyctl postgres db show jfl-platform-db

# Connect via psql
flyctl postgres connect -a jfl-platform-db

# List backups
flyctl volumes list -a jfl-platform-db

For migrations, use the app context:

# SSH into app machine and run migrations
flyctl ssh console -a jfl-platform -C "npx prisma migrate deploy"

Example output:

━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━
DATABASE - jfl-platform-db
━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━

Type: Postgres 15
Region: sjc (primary)
Storage: 10GB (2.3GB used)

CONNECTION
  Internal: jfl-platform-db.flycast
  Port: 5432

BACKUPS
  Last backup: 2h ago (automatic)
  Retention: 7 days

━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━
ACTIONS
━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━
/fly-deploy db connect     Open psql console
/fly-deploy db migrate     Run Prisma migrations

/fly-deploy ssh [machine-id]

SSH into a machine:

Options:

• No args - Pick from list of running machines
• machine-id - Connect to specific machine

Implementation:

# Interactive select
flyctl ssh console

# Specific machine
flyctl ssh console --machine 78d40e1a

# Run command
flyctl ssh console -C "ls -la"

Example:

Available machines:
1. 78d40e1a (sjc) - primary
2. 90e3f7a4 (sjc) - secondary

Which machine? (1-2, or 'all' to run command on all)

> 1

Connecting to 78d40e1a...

You're now in the machine. Type 'exit' to disconnect.

Common commands:
  ls /app              # App directory
  cat /app/package.json
  ps aux               # Running processes
  df -h                # Disk usage

/fly-deploy health

Check application health:

Checks:

• HTTP health endpoint
• All machines responsive
• Database connectivity
• Recent errors in logs

Implementation:

# Machine status
flyctl status --json

# Check health endpoint
curl https://jfl-platform.fly.dev/api/health

# Recent errors
flyctl logs --lines 100 | grep -i error

Example output:

━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━
HEALTH CHECK - jfl-platform
━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━

✓ HTTP health endpoint    200 OK (45ms)
✓ All machines running    2/2 healthy
✓ Database connected      Latency 12ms
⚠ Recent errors          3 errors in last hour

Recent issues:
  - 3x "Database connection timeout" at /api/projects

━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━

Want me to:
- Show full error logs?
- Check database connection pool settings?

/fly-deploy restart

Restart all machines:

Behavior:

1. Show current machine status
2. Confirm restart action
3. Rolling restart (one at a time for zero downtime)
4. Show new machine status

Implementation:

# Rolling restart
flyctl machines restart --force

# Or restart specific machine
flyctl machines restart 78d40e1a

Example:

This will restart all machines in a rolling fashion.

Current uptime: 3d 14h 22m

Continue? (yes/no)

> yes

Restarting machines...
  ✓ 78d40e1a restarted (15s)
  ✓ 90e3f7a4 restarted (12s)

All machines restarted. Health checks passing.

/fly-deploy destroy

Destroy the application and database:

DANGER ZONE

Behavior:

1. Explain what will be destroyed
2. Require explicit confirmation with app name
3. Optionally backup database first
4. Destroy resources
5. Update local files (remove fly.toml reference)

Implementation:

# List resources
flyctl apps list
flyctl postgres list

# Destroy app (requires confirmation)
flyctl apps destroy jfl-platform

# Destroy database (requires confirmation)
flyctl postgres destroy jfl-platform-db

Example:

⚠️  DANGER ZONE

This will permanently destroy:
  - App: jfl-platform (2 machines)
  - Database: jfl-platform-db (Postgres 15, 10GB)
  - All data (Users, Projects, Deployments, etc.)

This cannot be undone.

Create a database backup first? (yes/no)

> yes

Creating backup... ✓

To confirm, type the app name: jfl-platform

> jfl-platform

Destroying resources...
  ✓ App destroyed
  ✓ Database destroyed

Deployment removed from Fly.io.

Error Handling

Not Authenticated

If flyctl auth token fails:

You're not authenticated with Fly.io.

Run this to login:
  flyctl auth login

This will open your browser to authenticate.

App Not Found

If app doesn't exist on Fly:

No Fly.io app found.

Options:
1. Deploy new app: /fly-deploy init
2. Link existing app: flyctl config show [app-name]

Build Failures

If deployment build fails, parse logs and suggest fixes:

Build failed: Docker error

Error: "npm ERR! missing: typescript@^5.0.0"

Suggested fix:
  npm install --save-dev typescript@^5.0.0

Want me to:
- Install the missing dependency?
- Show full build logs?
- Cancel deployment?

Configuration Detection

Auto-detect from fly.toml

app = 'jfl-platform'
primary_region = 'sjc'

Read this to:

• Get app name
• Get primary region
• Get configuration

Auto-detect Database

# Check for attached Postgres
flyctl postgres list
flyctl postgres db show -a jfl-platform-db

Multi-App Detection

If multiple fly.toml files exist (monorepo):

Multiple Fly apps detected:

1. jfl-platform (./platform)
2. jfl-api (./api)
3. jfl-worker (./worker)

Which app? (1-3, or 'all')

Dependencies

• flyctl CLI installed and authenticated
• fly.toml in project directory
• Docker (for local builds)
• git (for commit detection)

Tips and Best Practices

Always show context

Don't just execute commands. Show what's happening:

Deploying changes from commit a4a4dfd...

Suggest next actions

After any operation, guide the user:

Deployed successfully!

Next steps:
- Test the deployment: https://jfl-platform.fly.dev
- Watch logs: /fly-deploy logs --follow
- Check health: /fly-deploy health

Warn on destructive actions

⚠️  This will restart all machines (brief downtime possible)

Parse errors helpfully

If something fails, explain what went wrong and how to fix it:

Deployment failed: Health check timeout

The app deployed but isn't responding to health checks.

Common causes:
1. Database connection issues (check DATABASE_URL secret)
2. Port mismatch (app listening on wrong port)
3. Startup errors (check logs)

Want me to:
- Check the logs?
- Verify database connection?
- Rollback to previous version?

Show costs when scaling

Scaling to 3 machines:
  Current cost: ~$5/mo
  New cost: ~$15/mo (+$10/mo)

Continue? (yes/no)

Integration with Other Skills

Works with /hud

Add deployment status to HUD:

Deployment: ✓ Production running (2 machines)
Last deploy: 2h ago

Works with project specs

Check product/platform/fly.toml for configuration.

Works with git

Detect uncommitted changes before deploy. Show commit info in deploy logs.

Quick Reference

/fly-deploy               # Status dashboard
/fly-deploy deploy        # Deploy latest changes
/fly-deploy logs -f       # Stream logs
/fly-deploy secrets list  # List secrets
/fly-deploy scale         # Scale resources
/fly-deploy db status     # Database info
/fly-deploy ssh           # SSH into machine
/fly-deploy health        # Health check
/fly-deploy restart       # Restart machines

---

Remember: Keep it conversational. Guide users through operations with clear feedback and helpful suggestions. Make Fly.io deployments feel simple and observable.