Skip to content
Agentic Coding Toolkit
This feature requires ACT Pro. Compare plans →

Priming

Load Flutter knowledge into the AI’s context before development. Priming gives the AI access to curated patterns, principles, and breaking changes so it writes Flutter code that follows best practices.

Commands

Section titled “Commands”

/act:prime-official-rules

Section titled “/act:prime-official-rules”

Load the official Flutter team coding guidelines.

Terminal window
/act:prime-official-rules

What it loads: The official Flutter/Dart style guide and coding rules from the Flutter repository.

When to use: When you want the AI to follow official Flutter conventions exactly.

/act:prime-patterns-lite

Section titled “/act:prime-patterns-lite”

Load essential patterns and principles (smaller context footprint).

Terminal window
/act:prime-patterns-lite

What it loads:

  • Critical principles (API key storage, avoiding global state, exception handling, YAGNI/KISS, etc.)
  • Breaking changes for Dart, Flutter, and Riverpod
  • Core patterns for common Flutter tasks

When to use: Default choice for most development work. Good balance of knowledge vs context usage.

/act:prime-patterns-full

Section titled “/act:prime-patterns-full”

Load the complete Flutter knowledge base.

Terminal window
/act:prime-patterns-full

What it loads: Everything in lite, plus:

  • All 23 patterns (column/row spacing, compute isolates, dot shorthand, enhanced enums, folder structure, theme extensions, etc.)
  • All 15 principles with full detail
  • Setup guides (Riverpod, flavors, environment variables, Sentry, etc.)

When to use: For complex features where you need deep pattern guidance, or when working with less common Flutter patterns.

Do I need to prime manually?

Section titled “Do I need to prime manually?”

Usually no. /act:workflow:work automatically loads Flutter knowledge based on the type parameter:

Terminal window
/act:workflow:work ai_specs/feature-plan.md        # loads lite (default)
/act:workflow:work ai_specs/feature-plan.md full    # loads full
/act:workflow:work ai_specs/feature-plan.md official # loads official rules only

Manual priming is useful when:

  • You’re working outside the standard workflow (e.g., free-form coding)
  • You want to prime before running /act:workflow:spec for better spec quality
  • You’re doing a code review and want Flutter best practices in context

What’s in the knowledge base?

Section titled “What’s in the knowledge base?”

Principles (critical standards)

Section titled “Principles (critical standards)”

Must-follow architectural guidelines:

  • API key storage — never hardcode secrets
  • Avoiding global state — use dependency injection
  • Avoiding tight coupling — depend on interfaces
  • Resource disposal — clean up controllers and subscriptions
  • Exception handling — visible errors, always log
  • YAGNI/KISS — keep it simple
  • Reactive state management — derive state, don’t duplicate it
  • Strongly-typed model classes — no raw maps or dynamic types
  • Widget classes over build helpers — composable, testable UI
Section titled “Patterns (recommended practices)”

Best practices and conventions for common Flutter tasks, including column/row spacing, compute isolates, constant sizes, dot shorthand syntax, enhanced enums, folder structure, logging, responsive design, semantics for accessibility, switch expressions, theme extensions, and more.

Breaking changes

Section titled “Breaking changes”

Documentation for breaking changes in Dart, Flutter, and Riverpod — so the AI doesn’t generate code that uses deprecated APIs.

Section titled “Related commands”