Flutter Claude Skills

A curated collection of Claude Code skills for Flutter / Dart development, testing, mobile app release workflows, and AI-assisted design-to-code pipelines.

Extracted and sanitized from real-world mobile app development experience.

Why this repo?
Quick Start
Start here — decision tree for your task
Playbooks — cross-skill composition patterns
What's inside
Installation
Skill anatomy
Sanitized placeholders
Contributing
License
Related

Why this repo?

Claude Code skills are reusable, triggerable knowledge packs that the agent loads when your task matches the trigger conditions. Building a solid skill library takes time — you have to discover the patterns, test them in real workflows, and iterate until the agent reliably picks the right one.

This repo gives you a production-tested starting point for Flutter-centric teams:

Real-world origin — every skill was extracted from active mobile app development, not invented in isolation.
Multi-layer coverage — testing, debugging, design-to-code, release, monetization, and collaboration workflows.
Sanitized & portable — secrets, personal paths, and company-specific references are stripped out (see Sanitized placeholders).
MIT licensed — drop it into your project, fork it, remix it.

If you're onboarding a team to Claude Code or want to skip months of trial-and-error discovering which skills actually matter for Flutter work, this is a reasonable baseline.

Quick Start

git clone https://github.com/ImL1s/flutter-claude-skills.git
cp -R flutter-claude-skills/skills/flutter-verify ~/.claude/skills/
# Open Claude Code in any Flutter project — flutter-verify auto-loads on trigger

That's it. Repeat cp -R for any other skills you want, or symlink the whole skills/ directory (see Installation).

Start here

Pick a row by your task, then follow the "Start with" skill and the linked playbook.

I want to...	Start with	Playbook	Category
ship a Flutter package / app to pub.dev or App Store	`writing-plans` → `flutter-verify`	#1 Pub package release cycle	Flutter core / Release
write a test-first feature or fix	`test-driven-development`	#2 TDD per task	Testing methodology
open / review a PR with rigor	`requesting-code-review` ↔ `receiving-code-review`	#3 Bidirectional code review	Code quality
debug an unfamiliar Flutter bug (crash, exception, hang)	`systematic-debugging`	#4 Debug Triage	Debugging
fix a UI layout / rendering / overflow bug (visual mismatch, golden fail)	`verify-ui` → `flutter-listview-viewport-gotchas`	—	Design / UI alignment
implement a design from Figma	`figma-use`	#5 Figma-to-Code Pipeline	Design / UI alignment
cut a store release (TestFlight / Play)	`release-preflight`	#6 Release Preflight Chain	Release / distribution
author a new Claude Code skill	`skill-creator`	#7 Skill Authoring Loop	Meta / skill authoring
initialize Firebase + monetization in a new app	`flutter-fullstack-init` or `firebase-flutter-setup`	#8 Firebase + Monetization Init	Firebase / Monetization
port native Android/iOS code to Flutter	`contract-based-porting`	—	Porting / refactoring
clean up AI-generated slop from a branch	`ai-slop-cleaner`	—	Code quality
drive Codex CLI from a script (success-check / diff-feed / worktree)	`codex-cli-rules`	—	Code quality
Figma MCP ran out of View seat quota	`figma-playwright-fallback`	—	Design / UI alignment
accidentally pushed a private email / API key to a public repo	`finishing-a-development-branch`	—	Git workflow
something else	Browse What's inside	—	—

Playbooks

Skills are most useful when composed. See PLAYBOOKS.md for eight cross-skill composition patterns distilled from real usage:

Pub package release cycle — writing-plans → flutter-verify
TDD per task — test-driven-development → testing-anti-patterns
Bidirectional code review — requesting-code-review ↔ receiving-code-review
Debug Triage — systematic-debugging → root-cause-tracing
Figma-to-Code Pipeline — figma-use → figma-implement-design → verify-ui
Release Preflight Chain — release-preflight → release-app
Skill Authoring Loop — skill-creator → writing-skills
Firebase + Monetization Init — flutter-fullstack-init or firebase-flutter-setup → admob-ux-best-practices

Each playbook names the skill chain, when it applies, and anchors to worked examples in PLAYBOOKS.md.

What's inside

67 skills organized by category. Each skill is a SKILL.md file that Claude Code auto-loads when its trigger conditions match.

Flutter core (14)

flutter-verify — post-code-change multi-layer verification (analyze + test + widget tree + runtime errors)
flutter-mcp-testing — orchestrate Marionette / Dart MCP / Mobile MCP / Codex for device testing
flutter-unit-testing — write effective Dart unit tests (Riverpod providers, freezed, etc.)
flutter-integration-testing — cross-module widget tests with real interactions
flutter-mobile-debugging — debug on real Android devices and iOS simulators
flutter-pub-get-stuck — diagnose hanging flutter pub get / build
flutter-listview-viewport-gotchas — recycling / viewport pitfalls in CustomScrollView
flutter-background-tasks — iOS BGTaskScheduler + Android WorkManager config
flutter-windows-ui-testing — automate Win32 testing for Flutter desktop
flutter-device-crud-testing — CRUD testing on real devices/simulators
flutter-api-reverse-engineering — Frida + BoringSSL traffic interception
flutter-social-login — Google / Apple Sign-In with Firebase Auth
flutter-fullstack-init — end-to-end project init with Firebase / AdMob / RevenueCat
fvm-flutter-release — fvm-based Flutter release workflow

Dart testing (3)

unit-testing-dart — write tests that actually catch bugs, not just satisfy coverage
integration-testing-dart — cross-module flow verification
api-contract-testing — Flutter/Dart API contract tests + reverse engineering

Firebase (4)

firebase-flutter-setup — Auth + AdMob + RevenueCat end-to-end
firebase-ai-logic — Gemini integration via Firebase AI Logic (formerly Vertex AI)
firebase-appcheck-manager — App Check attestation configuration
firebase-auth-manager — sign-in providers / authorized domains

Native / cross-platform (1)

kmp — Kotlin Multiplatform fundamentals (shared code, expect/actual, iOS integration, project setup)

Testing methodology (5)

test-driven-development — red/green/refactor loop enforcement
testing-anti-patterns — detect and avoid fake tests / mock overuse
verification-before-completion — require fresh evidence before claiming done
condition-based-waiting — replace flaky timing dependencies
webapp-testing — Playwright-based local webapp testing (web targets, not Flutter widgets)

Debugging (3)

systematic-debugging — four-phase diagnosis workflow
root-cause-tracing — trace errors back to original trigger
debug — Claude Code session / repo state diagnostics (not for Flutter app bugs — use systematic-debugging for those)

Design / UI alignment (8)

playwright-figma-scrape — systematic Figma scraping (screenshot / token / inventory) via Playwright MCP
figma-rest-api-scrape — faster/cleaner alternative via Figma REST API + PAT
figma-playwright-fallback — quick fallback when Figma MCP quota exhausts
figma-use — official Figma MCP prerequisite
figma-implement-design — translate Figma to production code
verify-ui — real-device UI comparison vs Figma
visual-verdict — structured screenshot comparison verdict
verify-ui-auto — ImageMagick SSIM / Golden Test automation

Release / distribution (6)

release-app — global TestFlight / Google Play release
release-preflight — pre-release checklist (build number / signing / metadata)
store-console-playbooks — App Store Connect + Google Play Console browser automation
mobile-store-upload-cli — iOS/Android CLI upload
store-screenshot-beautifier — store listing screenshot beautification
macos-notarization — macOS app notarization

Monetization (3)

admob-ux-best-practices — AdMob placement best practices
apple-appstore-manager — App Store Connect API management
revenuecat-manager — RevenueCat IAP configuration

Style — Anthropic-specific (1)

brand-guidelines — Anthropic brand colors / typography application (not Flutter-specific — applies to any Anthropic-branded artifact)

Collaboration (6)

receiving-code-review — how to process review feedback
requesting-code-review — how to request reviews
review-fix-commit — review → fix → commit cycle
subagent-driven-development — dispatch fresh-context subagents
dispatching-parallel-agents — parallelize independent failure investigation
testing-skills-with-subagents — validate skills against subagent pressure

Planning / skill authoring (7)

brainstorming — idea refinement before implementation
executing-plans — batch execution with review checkpoints
writing-plans — detailed plan authoring
writing-skills — skill authoring guidelines
skill-creator — create / modify / measure skills
deep-interview — Socratic requirements interview
deep-dive — trace + interview pipeline

Porting / refactoring (1)

contract-based-porting — zero-regression port via contract TDD

Code quality (3)

ai-slop-cleaner — regression-safe AI slop cleanup
codex-cli-rules — operational rules for driving Codex CLI from scripts (success-check / diff-feed / worktree / stdin)
defense-in-depth — multi-layer input validation strategy (architecture pattern, not crash debugging)

Git workflow (2)

using-git-worktrees — feature isolation with worktrees
finishing-a-development-branch — branch integration decisions

Installation

Drop any skill directory into your Claude Code skills path:

# User-global skills (available in all sessions)
cp -R skills/<skill-name> ~/.claude/skills/

# Project-local skills (only for one project)
cp -R skills/<skill-name> <your-project>/.claude/skills/

Or clone and symlink:

git clone https://github.com/ImL1s/flutter-claude-skills.git
ln -s "$(pwd)/flutter-claude-skills/skills/flutter-verify" ~/.claude/skills/flutter-verify

Claude Code auto-discovers skills on session start. Trigger conditions are declared in each SKILL.md frontmatter.

Skill anatomy

Each skill is a directory containing at minimum a SKILL.md with YAML frontmatter:

---
name: skill-name
description: |
  One-line summary + when to trigger.
  Triggers: keywords / conditions that activate this skill.
---

## Skill content
...

Some skills include helper scripts / templates / reference snippets in the same directory.

Sanitized placeholders

A few skills contain placeholder values you should replace for your project:

Skill	Placeholder	Replace with
`flutter-verify`	`/path/to/your/flutter/project`	Your actual Flutter project path
`figma-rest-api-scrape`	`YOUR_FIGMA_FILE_KEY`	Your Figma file key (from URL)
`figma-playwright-fallback`	`<your-figma-email>`	Your Figma account email

Contributing

PRs welcome. See CONTRIBUTING.md for detailed guidelines, or the short version:

Follow the SKILL.md frontmatter format
Include clear trigger conditions and when to skip
Add concrete examples with expected output
No hardcoded secrets / personal info / company-specific paths
Prefer Chinese or English (or both) — whichever serves clarity better

Also see:

Support

If this project saved you some time, you can buy me a coffee.

License

MIT — see LICENSE.

Originally authored by skill authors including Anthropic, oh-my-claudecode contributors, and this repo's maintainer. Each skill retains its original attribution in-file where applicable.

Claude Code — Anthropic's official CLI
oh-my-claudecode — multi-agent orchestration layer
Claude Code skills documentation

flutter-claude-skills