By mnfst-ai
Shape product specs into clean, validated briefs before AI coding: analyze docs/prompts for gaps, assumptions, risks, and drift; run user invalidation interviews; chunk tasks; craft guarded prompts; compare plans to git diffs for agentic tools like Claude Code.
Process what a builder heard in invalidation interviews. Score each assumption as validated, invalidated, or still open. Surface the pull signal if it appeared — lean-in, reach, referral, emotion, behavior, or time. Use this skill after running invalidation interviews, when a builder comes back from a conversation and says "here's what I heard," "did I find pull," or wants to know whether their assumptions survived contact with reality. Part of the Stage Manager Artful Making Skill Library by MNFST Ai.
Run a five-question readiness check before a builder crosses from shaping into staging. Use this skill when a vibe coder says "I think I'm ready to start building," wants to hand work to a coding tool but hasn't explicitly checked readiness, is feeling the pull to just start, or has been shaping for a while and wants to know if they're done. Also use when a builder is about to write their first prompt for a new story or feature. This skill prevents the two failure modes at the shape/stage boundary — premature staging and over-shaping. It should run before Chunking begins. Part of the Stage Manager Artful Making Skill Library by MNFST Ai.
Analyze a PRD, MRD, architecture doc, or prompt for specification gaps where coding tools will invent behavior on their own. Use this skill whenever a vibe coder or builder asks to "find the holes," "review my spec," "what will the AI fill in," "check my PRD," or wants to identify underspecified areas before handing work to Claude Code, Cursor, Replit, or any agentic coding tool. Part of the Stage Manager Artful Making Skill Library by MNFST Ai.
Analyze a PRD, MRD, architecture doc, prompt, or prototype for places where the builder has already committed to a typical solution without knowing it — especially when they are trying to push the envelope on something. Use this skill when a vibe coder asks "am I thinking too small," "are there better approaches," "what are my options before I build this," "am I moving too fast," or wants to protect a breakthrough idea from being averaged down by safe, default choices. Also use when transitioning from prototype to production. Part of the Stage Manager Artful Making Skill Library by MNFST Ai.
Analyze a PRD, MRD, architecture doc, design doc, or prompt for the load-bearing assumptions — the things that, if wrong, make everything downstream irrelevant. Use this skill when a vibe coder wants to know what to build first, asks "what could go wrong," wants to sequence work to surface the riskiest assumptions early, is worried about building the wrong thing, or wants to apply lean startup or Jobs to Be Done thinking before committing to a full build. Also use before writing prompts for Claude Code, Cursor, or Replit to ensure the highest-risk work gets proven before the lower-risk work gets built. Part of the Stage Manager Artful Making Skill Library by MNFST Ai.
Own this plugin?
Verify ownership to unlock analytics, metadata editing, and a verified badge. GitHub access is read-only (username + org membership).
Sign in to claimOwn this plugin?
Verify ownership to unlock analytics, metadata editing, and a verified badge. GitHub access is read-only (username + org membership).
Sign in to claimBased on adoption, maintenance, documentation, and repository signals. Not a security audit or endorsement.
How you move through your work is what you build.
A free, open collection of 16 skills for vibe coders who want to stay the author while building with AI tools.
These skills help you validate before building, expand options before converging, maintain coherence across a build, and keep the soul of the work intact while Claude Code, Cursor, Replit, or any agentic coding tool runs at speed.
claude plugin add --transport=github Mnfst-AI/Stage_Manager_Skills
That's it. All 16 skills appear as /slash-commands immediately.
git clone https://github.com/Mnfst-AI/Stage_Manager_Skills.git
cd Stage_Manager_Skills
bash install.sh
The install script symlinks each skill into ~/.claude/skills/ so they're available globally.
Start a session with /sm:stage-manage.
| Skill | Slash Command | What It Does |
|---|---|---|
| YAGNI | /sm:yagni | Gate: challenges the premise before a single line of code. "What did you find out, and who told you?" |
| Invalidation Prep | /sm:invalidate-prep | Excavates and ranks assumptions, identifies personas, generates problem statements. |
| Invalidation Interview | /sm:invalidate-interview | Generates a live interview script using negative framing to kill assumptions. |
| Invalidation Score | /sm:invalidate-score | Scores each assumption as validated, invalidated, or still open. Surfaces pull signals. |
| Skill | Slash Command | What It Does |
|---|---|---|
| Find the Holes | /sm:shape:find-holes | Maps every place a coding tool will invent behavior you didn't ask for. |
| Collapsed Options | /sm:shape:collapsed-options | Finds decisions you already made without knowing it. |
| Risk Sequence | /sm:shape:risk-sequence | Surfaces load-bearing assumptions and sequences by cost-if-wrong. |
| Soul Check | /sm:shape:soul-check | Deep read on whether the original animating idea is still alive. |
| Shape Brief | /sm:shape:brief | Synthesizes all shape skill output into ranked problems, inline change suggestions, and handoff documents. |
| Skill | Slash Command | What It Does |
|---|---|---|
| Shape-to-Stage Gate | /sm:shape:gate | Five readiness questions before staging begins. |
| Skill | Slash Command | What It Does |
|---|---|---|
| Chunking | /sm:stage:chunking | Breaks work into flow-cycle-sized pieces sequenced by cost of delay. |
| Prompt Craft | /sm:stage:prompt-craft | Turns a shaped chunk into a scoped, guardrailed prompt. |
| Live Mirror | /sm:stage:live-mirror | Compares plan vs. code output after a session. Surfaces invisible decisions. |
| Decision Capture | /sm:stage:decision-capture | Full decisions-made manifest after a build. |
| Skill | Slash Command | What It Does |
|---|---|---|
| Coherence Check | /sm:coherence-check | Lightweight 2-min gate at any transition point mid-build. |
| Stage Manage | /sm:stage-manage | Session opener and orchestrator. Watches for signals and suggests skills. |
/sm:yagni → /sm:invalidate-prep → /sm:invalidate-interview → /sm:invalidate-score → build
Shape skills (any combo) → /sm:shape:brief → Stage_Manager_Brief.md
/sm:stage-manage → watches for signals → suggests the right skill at the right moment
Start with /sm:stage-manage to open a session. It reads your project context and watches for signals — surfacing the right skill when it sees one.
The validation flow runs when you have a new idea. YAGNI asks the hard question. If you can't answer it, Invalidation Prep → Interview → Score walks you through finding out.
The Shape Brief flow runs any combination of shape skills, then synthesizes findings into ranked problems with inline change suggestions and handoff documents.
Coherence Check is the lightweight pulse check — any time something feels off mid-build.
Live Mirror catches invisible decisions per-session. Decision Capture catches everything across the full build.
.claude-plugin/plugin.json ← plugin manifest
install.sh ← manual installer
npx claudepluginhub mnfst-ai/stage_manager_skillsArtful Making skills for vibe coders. 11 skills across the Shape-to-Stage flow cycle — shape your spec, brief it, stage it clean. The mirror never goes dark.
Comprehensive skill pack with 66 specialized skills for full-stack developers: 12 language experts (Python, TypeScript, Go, Rust, C++, Swift, Kotlin, C#, PHP, Java, SQL, JavaScript), 10 backend frameworks, 6 frontend/mobile, plus infrastructure, DevOps, security, and testing. Features progressive disclosure architecture for 50% faster loading.
Develop, test, build, and deploy Godot 4.x games with Claude Code. Includes GdUnit4 testing, web/desktop exports, CI/CD pipelines, and deployment to Vercel/GitHub Pages/itch.io.
Comprehensive feature development workflow with specialized agents for codebase exploration, architecture design, and quality review
Design fluency for frontend development. 1 skill with 23 commands (/impeccable polish, /impeccable audit, /impeccable critique, etc.) and curated anti-pattern detection.
UI/UX design intelligence. 67 styles, 161 palettes, 57 font pairings, 25 charts, 15 stacks (React, Next.js, Vue, Svelte, Astro, SwiftUI, React Native, Flutter, Tailwind, shadcn/ui, Nuxt, Jetpack Compose). Actions: plan, build, create, design, implement, review, fix, improve, optimize, enhance, refactor, check UI/UX code. Projects: website, landing page, dashboard, admin panel, e-commerce, SaaS, portfolio, blog, mobile app. Elements: button, modal, navbar, sidebar, card, table, form, chart. Styles: glassmorphism, claymorphism, minimalism, brutalism, neumorphism, bento grid, dark mode, responsive, skeuomorphism, flat design. Topics: color palette, accessibility, animation, layout, typography, font pairing, spacing, hover, shadow, gradient.
Behavioral guidelines to reduce common LLM coding mistakes, derived from Andrej Karpathy's observations on LLM coding pitfalls