moinsen-dev

Set up a complete RevenueCat project from scratch — creates apps, products, entitlements, offerings, and packages in the correct order. Use when the user wants to create a new RevenueCat project, configure in-app purchases, set up subscriptions or…

Autonomy rules for EAS and store operations when an agent runs an Expo project unattended (YOLO mode) — what may run without asking, what always needs explicit approval, plus build-minute cost discipline. Use before any eas-cli, EAS-MCP build/workflow/submit,…

Use when the user asks to analyze or understand a RevenueCat experiment or its results.

Bootstrap a project with the moinsen local-only Expo workflow — creates AGENTS.md, STATE.md, docs skeleton (PRD/DESIGN/ARCHITECTURE/ROADMAP/AGENT-TOOLING) and .mcp.json from bundled templates. Use when starting a fresh Expo app, when the user asks to work…

Bootstrap a project with the moinsen Expo+Firebase workflow — creates AGENTS.md, STATE.md, docs (PRD/DEPLOY), planning/ task files and .mcp.json (scoped Firebase MCP) from bundled templates. Use when starting a fresh Expo app that uses Firebase, when the user…

Set up and troubleshoot the Expo MCP server for coding agents (live Expo docs, SDK-compatible package install, EAS builds/workflows, store data, local UI automation). Use when configuring MCP for an Expo project, when the Expo MCP connection fails, hangs, or…

Framework (OSS). Migrate an existing Apple/Swift Expo native module from the Expo Modules API 1.0 definition DSL to the 2.0 macro API (sometimes called v2) while preserving its JavaScript and TypeScript contract. Use when converting or incrementally adopting…

Run an Expo+Firebase app with strict dev/prod separation — two Firebase projects, environment switching via app.config.ts and EAS build profiles, and safe handling of google-services files. Use when setting up environments, adding Firebase config to an Expo…

Godot 4 GDScript idioms and the Godot 3 habits that still leak into generated code. Read this before writing or editing any .gd file. Covers signals, await, tweens, exports, typed GDScript, physics bodies, the renamed APIs that fail at parse time, and how to…

How to actually verify a Godot change instead of guessing. Read this before claiming any Godot work is done, and whenever a change needs to be seen running: launching the game, screenshotting it, injecting input, reading script errors with res:// line…

How to create and modify Godot scenes (.tscn) and resources (.tres) reliably — by writing a headless GDScript builder and letting Godot serialise the file, never by hand-writing scene text. Read this before creating any scene, instancing a scene into another,…

Designing what the game actually is, before and while building it — core loop, the verb, difficulty and pacing, progression and economy, session shape, and how to tell whether a design is working. Read this when starting a new game, when a game "has features…

Game feel — whether the avatar obeys the player. Read this when movement feels floaty, stiff, laggy or unfair, when tuning jumps, acceleration, input handling or camera follow, or before any polish pass. Covers input forgiveness, movement curves, response…

Juice — the layer of sensory feedback that makes a game feel alive without changing its rules. Read this when a game works but feels flat or dead, when adding impact to hits, shooting, pickups, UI or transitions, when doing a polish pass, or whenever "juice",…

Which Godot node to reach for, in 2D and in 3D, and how to structure a scene so it stays workable. Read this when starting a new scene, adding a character, camera, level geometry, UI or collision, or when unsure whether something should be an Area, a Body, a…

Runs the foundation council — five perspective agents (user-advocate, investor-advocate, architect, security-auditor, skeptic), a synthesizer, and a spine hydrator — before any feature or task work begins. Produces README.md as the project anchor,…

Initialises a project with the agentic-ai-features spine. Copies the platform-appropriate template into the current working directory (`CLAUDE.md` for Claude Code, `AGENTS.md` for Codex), fills in basic placeholders from the repo, and stops without…

Audits whether planned or claimed work is supported by current repository evidence. Use for status sweeps, task completion checks, or before handoff. It does not write feature code.

Creates a decision-complete plan for a single feature, refactor, or task bundle in this repository. Use before implementation when the requested work is larger than a direct one-file change or requires product, API, schema, dependency, or workflow decisions.

Implements exactly one scoped task from this repository's plan. Use when the user asks to implement a specific TASK or the next approved task. Stops after the task is verified and reviewed; does not continue through the plan by default.

Walks a multi-task plan file autonomously. Per task it dispatches implementer + verifier + reviewer as separate agents, commits, then advances. Stops at human gates, repeated failures, or end-of-plan. Use when the user asks to "walk the plan", "run the next N…

Perform an independent adversarial review of an integrated game or software project using reproducible runtime, visual, interaction, accessibility, determinism, and performance evidence. Use after a Project Forge slice enters review, before accepting a…

Bootstrap an original game or software project from a concept by producing a product contract, architecture, ownership plan, quality gates, evidence plan, and repository scaffold. Use when starting a new project, turning an idea into an executable build plan,…

Run Claude Code as an endless daemon with inbox/outbox task queue. Drop tasks into inbox, collect results from outbox. Fresh session per task prevents context rot. Use for permanent Claude agents, job runners, drop-folder workflows, or task queues.

Run Claude Code autonomously overnight with a validated runbook, security hooks, heartbeat watchdog, and kernel sandbox. Use when someone wants unattended Claude runs, headless batch work, or autonomous overnight coding.

Use when working with Flutter apps that use moinsen_runapp — for setup, configuration, state exposure, error reporting, HTTP monitoring, UI interaction (tap, scroll, text input), MCP server setup, and LLM-assisted debugging via CLI, VM Service extensions, or…

Use when working with Flutter apps that use moinsen_runapp — for setup, configuration, state exposure, error reporting, HTTP monitoring, UI interaction (tap, scroll, text input), MCP server setup, and LLM-assisted debugging via CLI, VM Service extensions, or…

Einem Wissensverbund aus mehreren Hermes-Agenten beitreten und ihn bedienen — geteiltes Gedächtnis über ein privates Git-Repo, verschlüsseltes Vault (sops+age), signaturgeprüfte Nachrichten zwischen Agenten, nur signierte Releases. Nutze diesen Skill, wenn du…

Use vibe-secrets (local encrypted vault) to populate .env files and manage API keys in a project. Trigger when the user mentions a missing env var, asks to set up a new project with keys, asks how to rotate/inject secrets, when you see an .env.example but no…

Generate devhub.toml configuration files and manage multi-project development environments. Use when setting up projects, creating service configs, starting/stopping services, checking status, or managing development workflows. Triggers on mentions of devhub,…

Inspect, diagnose, plan, and safely operate Xcode, Flutter, and Expo app workspaces through RunDocket's framework-neutral CLI or MCP contract. Use when an AI agent needs to identify the owning app in a repository, check local toolchain or device readiness,…

Start and stop a native macOS screen recording of a specific window (not the whole screen) from the command line. Use when a user wants to record an app window, a demo, a bug repro, or a test run on a Mac, and wants it driven programmatically by an agent…

Evidence-gated problem validation, Lean Startup validated-learning experiment design, Lean sparring-partner critique, and opportunity shaping for product, software, service, and non-software ideas. Use when a user wants to capture, park, validate, challenge,…