Skip to content
Agentic Coding Toolkit

Quickstart

Get from zero to a completed spec, plan, and working implementation in under 10 minutes.

Make sure you’ve completed the Installation steps first.

1. Create a spec

Section titled “1. Create a spec”

Navigate to your Flutter project (or create a new one) and start a Claude session:

Terminal window
cd my_flutter_app
claude

Describe a feature you want to build:

Terminal window
/act-workflow-spec "add a settings screen with dark mode toggle and language selection"

ACT will:

  • Ask 2-4 clarifying questions to fill gaps
  • Show you a preview outline for approval
  • Generate a detailed specification with user flows, requirements, and edge cases
  • Save it to ai_specs/settings-spec.md

Review the spec. This is your chance to catch problems before any code is written.

2. Create a plan

Section titled “2. Create a plan”

Turn the spec into a phased implementation plan:

Terminal window
/act-workflow-plan ai_specs/settings-spec.md

ACT analyzes your codebase for conventions and patterns, then creates a plan with:

  • Phases broken into concrete tasks with file paths
  • TDD test-first tasks where applicable
  • Verification steps for each phase

The plan is saved to ai_specs/settings-plan.md.

3. Build it

Section titled “3. Build it”

Execute the plan:

Terminal window
/act-workflow-work ai_specs/settings-plan.md

ACT will:

  • Load Flutter knowledge (patterns, principles, breaking changes)
  • Execute each phase in order
  • Reconcile plan checkboxes and blocker notes after each completed phase
  • Run flutter analyze and flutter test before each phase is considered done
  • Commit successful completed phases by default; pass --do-not-commit to keep changes local
  • Pass --create-pr to push and open a PR after the full plan succeeds

What just happened?

Section titled “What just happened?”

You completed the full ACT workflow:

  1. Spec — defined what to build with clarity
  2. Plan — broke it into phases with concrete tasks
  3. Work — executed the plan with automated verification

The key insight: by spending a few minutes on the spec, you caught ambiguity before implementation. The AI had clear instructions, which means fewer bugs, less rework, and a better result.

Next steps

Section titled “Next steps”