Agent skills for mobile app work, with a focus on Flutter.
Each skill is a directory under skills/ with a SKILL.md in the Agent Skills format.
Many skills adapt someone else's video, article, or post, and credit the original in a Sources section.
Skills that depend on libraries, SDKs, or store policies record in metadata.docs-verified the date they were last checked against official documentation, and list those docs under Sources.
Claude Code loads skills from ~/.claude/skills/ for all projects and from .claude/skills/ for a single project.
git clone https://github.com/mikezamayias/agent-skills.git
mkdir -p ~/.claude/skills
# One skill
cp -R agent-skills/skills/launch-readiness ~/.claude/skills/
# Every skill, as symlinks that pick up `git pull`
for d in "$PWD"/agent-skills/skills/*/; do ln -s "${d%/}" ~/.claude/skills/; doneOther agents that read Agent Skills folders take the same directories. Copy or symlink each skill folder into the agent's skills directory.
| Agent | Skills directory |
|---|---|
| Codex | ~/.codex/skills/ |
| Cursor | ~/.cursor/skills/ |
| Pi | ~/.pi/agent/skills/ |
These paths change between agent versions, so check your agent's documentation.
Some skills call external tools, such as the Codex CLI, the asc App Store Connect CLI, Sentry, or mobile-mcp.
Each SKILL.md states what it needs.
uv run --with pyyaml python3 toolbox/validate-skills.pyThe validator checks names, descriptions (at most 1024 characters and 3 sentences), frontmatter keys, the docs-verified date format, and local links.
Markdown is formatted with Prettier and linted with markdownlint, the same tools the VS Code extensions esbenp.prettier-vscode and davidanson.vscode-markdownlint run.
Both CLIs are pinned to the versions those extensions bundle, and .markdownlint-cli2.jsonc holds the lint config for both the CLI and the extension.
npx [email protected] --check "**/*.md"
npx [email protected] "**/*.md"These are skills I wrote or adapted. Skills that adapt someone else's work credit it in the skill.
| Skill | What it does |
|---|---|
| app-graphics | Generates store icons, feature graphics, social banners, and a press hero from a project's local files. |
| app-landing-page | Generates five single-file landing pages, each with a different design philosophy. |
| app-screenshots | Orchestrates App Store and Play Store screenshot capture, framing, and locale fan-out for a Flutter app. |
| break-ui | Stress-tests a Flutter screen with worst-case data, text scale, and RTL, and leaves a widget test behind. |
| consumer-app-from-behavior | Turns observed consumer behavior into a core loop, a reference teardown, and an MVP spec. |
| cross-language-safety | Prevents type mismatches between PostgreSQL, PostgREST, Deno, JSON, and Dart. |
| dart-tool-scripts | Moves non-trivial CI and bootstrap logic from Bash into tool/*.dart scripts. |
| flutter-app-size-audit | Cuts Flutter install size through build flags, asset formats, and shrinking. |
| flutter-async-at-the-edge | Keeps async work in services and repositories so views render synchronous sealed state. |
| flutter-async-cubit-safety | Guards every emit after an await to prevent emit-after-close crashes. |
| flutter-bloc-clean-architecture | Applies BLoC/Cubit with Very Good Ventures conventions on Feature-First Clean Architecture packages. |
| flutter-clean-architecture-tdd | Builds a Flutter feature test-first across its domain, data, and presentation packages. |
| flutter-cubit-smart-caching | Adds time-based freshness checks so cubits skip redundant reloads on navigation. |
| flutter-error-logging-conventions | Replaces silent catch (_) blocks with consistently prefixed logging. |
| flutter-observability-instrumentation | Adds type-safe analytics, tracing, metrics, and breadcrumbs with Sentry and PostHog. |
| flutter-paywall-and-review-timing | Covers IAP wiring, App Store review prompt limits, and a decaying paywall schedule. |
| flutter-precommit-cleanup | Formats, analyzes, regenerates code, and squashes WIP commits before a PR. |
| flutter-router-decision | Picks a Flutter router and keeps go_router unless there is a concrete need to switch. |
| flutter-runtime-safety | Catches errors that escape try-catch and stops restored backups from bringing back bad state. |
| flutter-shorebird-ota | Ships Dart-only over-the-air patches with Shorebird, including in CI. |
| flutter-stream-architecture | Multicasts repository streams and limits Bloc rebuilds. |
| flutter-strict-analysis | Adopts a strict lint policy and enforces it in CI with --fatal-infos. |
| flutter-web-first-paint | Fixes blank first loads on Flutter web with an HTML splash, deferred imports, and preloading. |
| forced-pivot-15-day | Sequences a one to three week pivot forced by a store guideline, policy, or deadline. |
| gplay-console | Preflights and automates Google Play releases, tracks, store listings, Data safety, and vitals. |
| ios-distribution-feature-flags | Sets compile-time feature flags for Debug, TestFlight, and App Store builds. |
| launch-readiness | Runs a tiered Flutter launch audit and produces a scored SHIP, HOLD, or BLOCK report. |
| launch-watch | Proves launch telemetry arrives, then watches analytics, crashes, email, and feedback for three weeks. |
| market-pulse | Produces an app market analysis covering competitors, sizing, pricing, and projections. |
| mobile-ui-reference | Designs Flutter screens from shipped references instead of generated UI. |
| paywall-conversion-flow | Orders onboarding, the rating prompt, and the paywall so value comes first. |
| paywall-transparency | Shows users what they get and when they will be charged before the paywall. |
| pr-preview-shipping | Makes every pull request produce an installable preview. |
| proactive-crash-repair-loop | Turns Sentry crash telemetry into deduplicated, human-reviewed draft fix PRs. |
| promise-audit | Traces every promise in a privacy policy, terms, or marketing claim to the mechanism behind it. |
| ship-first-dollar-app | Validates paid pain, scopes a one-feature MVP, and measures the path to first revenue. |
| slim-ios-simulator-fleet | Fits more iOS simulators on one Mac with simslim. |
| store-launch-mastermind | Orchestrates App Store and Google Play go-live into one go/no-go report. |
| Skill | What it does |
|---|---|
| academic-research-assistant | Synthesizes several research papers into one literature review of claims, debates, and consensus. |
| answer-sheet | Puts open questions on one self-contained HTML page and saves the answers to a dated Markdown file. |
| decision-sweep | Logs each ruling once and updates every dependent document, then runs checks. |
| engineering-diary | Keeps a daily engineering diary so a solo developer does not lose context between sessions. |
| figma-review-loop | Runs a Figma review round: reads comments, records rulings, redesigns frames, and sweeps instances. |
| moscow-prioritization | Cuts a noisy backlog with Must, Should, Could, and Won't. |
| orchestrate-agentic-engineering | Coordinates one human, one orchestrator, and bounded specialist agents from brief to end-to-end proof. |
| parallel-agent-inbox | Runs many coding-agent threads as an inbox and only pulls the human in when work is done or blocked. |
| spec-sql-harness | Proves SQL from design documents by loading it into the production Postgres image with smoke tests. |
| sync-context | Builds a morning briefing from the last 7 days of GitHub, Slack, and PostHog activity. |
| task-blitz | Pulls open tasks from Notion and runs independent ones with parallel sub-agents that plan before coding. |
| techdebt | Scans a project for duplicate code, dead code, TODOs, and unused imports, and writes a prioritized report. |
| vendor-decision | Chooses or re-checks a backend, API, or service from live evidence, competitors, and gatekeepers. |
| web-interface-review | Reviews web UI code for accessibility, focus, forms, dark mode, i18n, and phone behavior. |
I also use these skills as published by their authors. They are not copied here, so install them from their source.
| Source | Skills |
|---|---|
| clawhub.ai/0xterrybit/skills/todo | todo |
| clawhub.ai/bjesuiter/skills/prd | prd |
| clawhub.ai/chandrasekar-r/skills/security-audit | security-audit |
| clawhub.ai/julianengel/skills/remind-me | remind-me |
| clawhub.ai/michaelgathara/skills/youtube-watcher | youtube-watcher |
| clawhub.ai/pors/skills/research | research |
| clawhub.ai/zats/skills/perplexity | perplexity |
| Appllama/appllama-skills | appllama-app-design-skill, appllama-usage |
| besoeasy/open-skills | humanizer |
| clawdbot/linkedin-cli | linkedin-cli |
| cloudflare/skills | agents-sdk, durable-objects, turnstile-spin, web-perf |
| codeswithroh/tastemaker | tastemaker |
| dietrichgebert/ponytail | ponytail |
| herdrdev/herdr | herdr |
| heygen-com/hyperframes | faceless-explainer, hyperframes, hyperframes-animation, hyperframes-audio, hyperframes-cli, hyperframes-core, hyperframes-creative, hyperframes-keyframes, hyperframes-registry, hyperframes-studio, media-use |
| multica-ai/andrej-karpathy-skills | karpathy-guidelines |
| openclaw/openclaw | apple-notes, apple-reminders, coding-agent, gemini, peekaboo, skill-creator |
| pskoett/self-improving-agent | self-improving-agent |
| rorkai/app-store-connect-cli-skills | asc-app-create-ui, asc-aso-audit, asc-build-lifecycle, asc-cli-usage, asc-crash-triage, asc-id-resolver, asc-localize-metadata, asc-metadata-sync, asc-notarization, asc-ppp-pricing, asc-release-flow, asc-revenuecat-catalog-sync, asc-screenshot-resize, asc-shots-pipeline, asc-signing-setup, asc-submission-health, asc-subscription-localization, asc-testflight-orchestration, asc-wall-submit, asc-whats-new-writer, asc-workflow, asc-xcode-build |
| twostraws/Swift-Concurrency-Agent-Skill | swift-concurrency-pro |
| twostraws/Swift-Testing-Agent-Skill | swift-testing-pro |
| twostraws/SwiftUI-Agent-Skill | swiftui-pro |
| vercel-labs/agent-browser | agent-browser, agent-browser-legacy |
| VeryGoodOpenSource/vgv-ai-flutter-plugin | vgv-accessibility, vgv-dart-flutter-sdk-upgrade, vgv-green-gate, vgv-license-compliance, vgv-static-security, vgv-testing, vgv-ui-package, vgv-very-good-analysis-upgrade |
| VeryGoodOpenSource/vgv-ffca-plugin | ffca-architecture, ffca-audit, ffca-cross-feature, ffca-feature, ffca-routing |
Open an issue for changes. This repository is generated, so pull requests would be overwritten.
BSD 3-Clause. See LICENSE.