vp-k/flutter-craft/flutter-parallel-agents

Trust tier 3
  • community
Source check pending

Checking that the SKILL.md file is still in the repository.

Use when facing 2+ independent Flutter tasks that can be worked on without shared state or sequential dependencies

Install

npx vskill@latest install vp-k/flutter-craft/flutter-parallel-agents

SKILL.md

Dispatching Parallel Flutter Agents

Overview

When you have multiple unrelated tasks (different features, different test files, different bugs), working on them sequentially wastes time. Each investigation is independent and can happen in parallel.

Core principle: Dispatch one agent per independent problem domain. Let them work concurrently.

Announce at start: "I'm using the flutter-parallel-agents skill to handle these tasks."

When to Use

Multiple tasks/failures?
├─ NO → Single agent handles it
└─ YES → Are they independent?
├─ NO (related) → Single agent investigates all
└─ YES → Can they work in parallel?
├─ NO (shared state/files) → Sequential agents
└─ YES → Parallel dispatch (this skill)

Use when:

  • Multiple test files failing with different root causes
  • Multiple features to implement in different areas
  • Each problem can be understood without context from others
  • No shared state between investigations
  • Different Clean Architecture layers to work on

Don't use when:

  • Failures are related (fix one might fix others)
  • Tasks modify same files
  • Tasks have dependencies on each other
  • Need to understand full system state first

Flutter-Specific Parallel Scenarios

Independent Layer Tasks
lib/features/auth/
├── domain/ ← Agent 1: Implement entities & interfaces
├── data/ ← Agent 2: Implement models & datasources
└── presentation/ ← Wait for Agents 1 & 2 (has dependencies)

Can parallelize: Domain + Data layers (different files) Cannot parallelize: Presentation depends on Domain

Independent Feature Tasks
lib/features/
├── auth/ ← Agent 1: Auth feature
├── profile/ ← Agent 2: Profile feature
└── settings/ ← Agent 3: Settings feature

Can parallelize: All three (different feature folders)

Independent Test Fixes
test/features/
├── auth/data/repositories/auth_repo_test.dart ← Agent 1
├── profile/presentation/bloc/profile_bloc_test.dart ← Agent 2
└── settings/data/datasources/settings_ds_test.dart ← Agent 3

Can parallelize: Different test files, different features

The Pattern

1. Identify Independent Domains

Group tasks by what's affected:

  • Feature A: User authentication
  • Feature B: Profile management
  • Feature C: Settings storage

Each domain is independent - fixing auth doesn't affect settings.

2. Create Focused Agent Tasks

Each agent gets:

  • Specific scope: One feature or test area
  • Clear goal: What to implement/fix
  • Constraints: Don't change other features
  • Expected output: Summary of what was done
3. Dispatch in Parallel

Use subagent_type: "flutter-craft:flutter-dart-specialist" for implementation agents — its definition carries the Flutter/Dart norms (Riverpod patterns, widget principles, recommended packages), so prompts stay task-only.

Task("Implement auth domain layer - entities and repository interface")
Task("Implement profile domain layer - entities and repository interface")
Task("Implement settings domain layer - entities and repository interface")
// All three run concurrently, each as flutter-craft:flutter-dart-specialist
4. Review and Integrate

When agents return:

  • Read each summary
  • Verify fixes don't conflict
  • Run flutter analyze
  • Run flutter test
  • Integrate all changes

Agent Prompt Structure

markdown
Implement the [Layer] layer for [Feature] feature.
## Scope
lib/features/[feature]/[layer]/
## Files to Create
-entities/[entity].dart
-repositories/[repository].dart (interface)
## Requirements
[Paste specific requirements from plan]
## Constraints
-ONLY modify files in lib/features/[feature]/[layer]/
-Do NOT touch other features
-Follow Clean Architecture principles
-Run flutter analyze before committing
## Expected Output
-Summary of what you created
-flutter analyze result
-Commit message used

Example: Parallel Feature Implementation

Scenario: Need to implement domain layers for 3 features

Decision: Each domain layer is independent

Dispatch:

Agent 1 → Implement auth/domain/ (User entity, AuthRepository interface)
Agent 2 → Implement profile/domain/ (Profile entity, ProfileRepository interface)
Agent 3 → Implement settings/domain/ (Settings entity, SettingsRepository interface)

Results:

  • Agent 1: Created User entity, AuthRepository ✅
  • Agent 2: Created Profile entity, ProfileRepository ✅
  • Agent 3: Created Settings entity, SettingsRepository ✅

Integration:

bash
flutter analyze # All clean
flutter test # All pass
git log --oneline -3 # See all three commits

Common Mistakes

❌ Too broad: "Implement all features" - agent gets lost ✅ Specific: "Implement auth/domain layer" - focused scope

❌ Dependent tasks in parallel: Data layer before Domain ✅ Correct order: Domain first, then Data in parallel

❌ Same files: Two agents editing pubspec.yaml ✅ No conflicts: Each agent has isolated file scope

When NOT to Use

  • Sequential dependencies: Data layer needs Domain layer first
  • Shared files: Multiple agents need same file
  • Cross-feature dependencies: Feature B imports from Feature A
  • Need global context: Understanding requires seeing entire app

Verification

After agents return:

  1. Review each summary - Understand what changed
  2. Check for conflicts - Did agents edit same files?
  3. Run flutter analyze - Verify all code is clean
  4. Run flutter test - Verify tests pass
  5. Check imports - No circular dependencies
bash
flutter analyze
flutter test
git diff --stat # See all changes

Benefits

  1. Parallelization - Multiple implementations simultaneously
  2. Focus - Each agent has narrow scope
  3. Independence - Agents don't interfere
  4. Speed - N problems solved in time of 1

Security

  • Security scanPassed, 95 of 100Pattern checks on the skill's files.
  • Model reviewPassed

Quality evaluation

  • Task evaluationNot runNo evaluation yet. Run one in Skill Studio.

Source

Popularity

Repository stars
4
Repository forks
2
7-day trend score
8

Stars and forks are for the whole GitHub repository, not this skill alone. Refreshed 17 Mar 2026, 7 months ago.

Works with

Any agent vskill installs to: 56 targets in all. These 39 have a page on this site.

  • Claude Code
  • Cursor
  • GitHub Copilot
  • Windsurf
  • Codex
  • Gemini CLI
  • Cline
  • Amp
  • Roo Code
  • Goose
  • Continue
  • Kilo Code
  • Junie
  • Trae
  • OpenHands
  • Qwen Code
  • AI Agent
  • Tabnine Chat
  • Sourcegraph Cody
  • Amazon Q
  • Firebase Genkit
  • Sweep
  • Devin
  • Phind
  • Mentat
  • Claude Desktop
  • Aider
  • GPT Pilot
  • Plandex
  • bolt.new
  • v0
  • Replit Agent
  • Anse
  • TypingMind
  • BetterChatGPT
  • LibreChat
  • Warp AI
  • Pieces
  • Fig AI
Embed badge
vskill VERIFIED badge
[![vskill VERIFIED](https://verified-skill.com/api/v1/skills/vp-k/flutter-craft/flutter-parallel-agents/badge)](https://verified-skill.com/skills/vp-k/flutter-craft/flutter-parallel-agents)

Something wrong with this skill or its page? Tell us.