Agent 06 — Deployer

Deployer

From working code to a running production environment.

Building the app is only half the job. Deployer plans the complete path to production — and starts by telling you exactly how to work through your Spec4 phases with the AI coding agent you're already using. For cloud deployments, Deployer can even generate complete Terraform scripts.

Two things Deployer does

Deployer works in two distinct parts. The first gets you coding immediately. The second ensures you have somewhere solid to ship to.

Part 1

Coding Agent Guidance

You pick your AI coding agent — Claude Code, Cursor, GitHub Copilot, Windsurf, Cline, or others. Deployer searches for current documentation on that agent and gives you precise instructions for how to use it with your Spec4 phases: where the files live, how to load them, and agent-specific tips to avoid common pitfalls.

Part 2

Deployment Planning

Seven focused questions — one per turn — covering your deployment target, containerization, CI/CD pipeline, environment configuration, monitoring, and observability. For cloud deployments, Deployer can generate complete Terraform infrastructure scripts from scratch.

Coding agent guidance

Every AI coding agent has its own workflow for loading files, running commands, and verifying phases. Deployer looks up current documentation and gives you agent-specific guidance for:

Getting started

How to open and configure your agent in your project directory, with any flags or settings that improve reliability.

Loading phase files

The exact syntax for referencing .spec4/phases/phase1.json with your specific agent — commands, flags, and file-reference patterns.

Recommended workflow

The phase-by-phase approach: implement, run the verification command, confirm it passes, then move to the next phase. Don't skip ahead.

Agent-specific caveats

Known pitfalls, context limits, and behavior patterns specific to the agent you chose — so you don't run into surprises mid-implementation.

Deployment planning

Seven focused questions, one per turn. Deployer uses web search to verify current pricing, free-tier availability, and setup documentation before making recommendations.

Target type

Cloud provider (AWS, GCP, Azure, DigitalOcean), PaaS (Fly.io, Render, Railway, Vercel, Netlify), on-premise, or serverless.

Specific service

For cloud: ECS vs. App Runner vs. Lambda. For PaaS: Fly.io vs. Render vs. Railway. Honest trade-off analysis for your specific project.

Containerization

Docker setup with currently recommended base images, multi-stage builds, and container registry configuration.

CI/CD pipeline

GitHub Actions, GitLab CI, CircleCI, Bitbucket Pipelines — with trigger branch, stages, and deployment steps configured for your target.

Environment config

Required environment variables, secrets management strategy (Vault, AWS Secrets Manager, environment files), and configuration patterns.

Monitoring

Error tracking (Sentry, Rollbar) and metrics (CloudWatch, Datadog, Grafana Cloud) — configured and integrated from day one, not bolted on later.

Terraform (cloud deployments only)

For cloud targets, Deployer can generate complete, production-ready Terraform scripts — all .tf files needed to provision your infrastructure from scratch. No partial templates. No placeholders.

What Deployer produces

Deployer saves a complete deployment-plan.md to your .spec4/ directory. The document includes:

→ Coding Agent Guidance — how to use your chosen agent with Spec4 phase files, step by step

→ Target configuration — provider, service, region, and hosting approach

→ Containerization — base image, registry, and build configuration

→ CI/CD pipeline — platform, trigger branch, and stage configuration

→ Ordered deployment steps — exact shell commands in sequence, from local to live

→ Complete configuration files — ready-to-use Dockerfile, CI/CD YAML, and any other files needed

→ Terraform scripts (if requested) — all .tf files for full infrastructure provisioning from scratch

→ Notes — cost estimates, caveats, and additional advice specific to your setup

When to use Deployer

✓ After Phaser, before you start implementing — to know how you'll ship before you start building

✓ When you want to understand how to use Spec4 phases with your specific AI coding agent

✓ When you're planning infrastructure from scratch and want a complete, validated plan

✓ When you want CI/CD and monitoring configured from day one rather than added later