Viral Product Evaluator by luongnv89 — AI Skill for Any LLM

Overview

Publisher	luongnv89
Repository	`skills`
Skill name	`viral-product-evaluator`
Stars	124
Forks	18
Bundled files	4
License	MIT
Links	Skill source Repository Homepage

Markdown instructions

A SKILL.md file the model loads on demand, so it only costs tokens when a request actually matches.

Works with any LLM

AI skills are plain Markdown, not provider-specific code, so this works with GPT, Claude, Gemini, Grok, or a local model.

4 bundled files

Scripts, templates, and references the model can read while it works. Files are read-only and never executed.

Open source

Published by luongnv89 on GitHub. Read the source before you install it.

Installation

Install the Viral Product Evaluator AI skill in TypingMind to use it with any LLM, or drop it into another agent that reads SKILL.md.

Install in TypingMind

TypingMind installs a skill straight from its GitHub folder — it reads SKILL.md, bundles the resource files, and stores the result locally.

Open the app and go to Plugins → Skills.
Choose "Install from GitHub".
Paste the skill folder URL below and confirm.
Enable the skill in any chat where you want it available.

Plugins → Skills → Add skill → From GitHub URL, then paste the folder URL and press Continue.

Install in another agent

Any agent that reads the Agent Skills format can use this skill — copy the folder into that agent's skills directory.

Claude Code — .claude/skills

git clone --depth 1 https://github.com/luongnv89/skills.git /tmp/skills
mkdir -p .claude/skills
cp -r /tmp/skills/skills/viral-product-evaluator .claude/skills/viral-product-evaluator

Restart Claude Code after copying so it picks up the new skill.

Use it in TypingMind

Enable Viral Product Evaluator in any TypingMind chat and the model takes it from there. Its name and description sit in the system prompt, and the moment a request matches, the model loads the full instructions itself — you never invoke it by hand, and it costs no tokens until it is actually used.

The model loads Viral Product Evaluator on its own as soon as a request matches it.

Works with any AI model

AI skills are plain Markdown instructions rather than provider-specific code, so Viral Product Evaluator is not tied to the model it was written for. Install it once in TypingMind and use it with GPT-5, Claude, Gemini, Grok, DeepSeek, Mistral, Llama, or a local model you run yourself — all on your own API keys.

Loaded only when it is needed

The system prompt carries just the name and description. The instructions are fetched on the first matching request, so an idle skill costs nothing.

Switch models mid-chat

Because the skill is instructions rather than code, changing model does not break it — the next model reads the same SKILL.md.

Skill instructions

This is the SKILL.md content the model loads. Read it before installing — a skill is instructions your model will follow.

Viral Product Evaluator

Grade a product against the 32 principles of viral products. Two inputs — a codebase and a landing page — produce one output: a scored report of what's already satisfied and, in priority order, what to do next to make it more viral.

When to Use

Trigger when the user wants to:

Make a product, SaaS, or indie app "more viral" or more shareable
Score / audit a landing page against viral-marketing or conversion principles
Get a prioritized, ordered list of changes to improve a product's pull
Check a product against "the 32 principles" (Marc Lou-style viral-product rules)

Do not use for: technical SEO (seo-ai-optimizer), App Store ASO (aso-marketing), turning a README into a page (landing-page-generator), or bug-hunting code review (code-review). This skill evaluates and prioritizes; it does not rewrite the product.

Prerequisites

Read access to the target codebase directory.
Landing page signal: public URL, local file, or auto-detectable in the tree.
The skill's references/*.md files present for the rubric and output shape.
User confirmation on any deviation from the 32 principles.

Missing prerequisites → stop and report before gathering evidence.

What this skill does and does not touch

It reads the codebase, fetches the landing page, and writes one report file (viral-evaluation.md). It does not edit the user's source, change copy, or commit anything — so no repo-sync/branch guardrail is needed. If the user then asks you to apply fixes, that is a separate task (hand off to a copy/frontend skill); this skill stops at the prioritized plan.

Dependency Preflight (mandatory)

This skill invokes /browse, and only on the live URL path — a local file or an auto-detected page needs nothing installed. Resolve it before fetching anything:

bash
test -d "$HOME/.claude/skills/browse" || asm list -p claude --json | grep -q '"browse"' || {
  echo "Missing required skill: browse" >&2
  echo "Install it:      asm install github:garrytan/gstack:browse -p claude -s global --yes" >&2
  echo "No asm yet:      npm install -g agent-skill-manager" >&2
  echo "Verify:          asm list -p claude --json | grep 'browse'" >&2
}

The install-path test runs first because /browse ships in gstack and may be present without asm knowing about it. -p claude -s global is required: asm install will not guess a provider non-interactively, and a project-scoped install lands where the $HOME test cannot see it.

A missing /browse is fail-soft, not fatal: print the commands above, then ask the user for a local file or saved HTML of the page. Never score a URL you could not load — do not invent one.

Inputs

Accept any combination the user provides; ask only for what's missing and truly needed.

Landing page — one of:
- a live URL → fetch it with the /browse skill (headless). Capture rendered copy, headline, CTAs, pricing section, testimonials, nav, and <head> meta (og:image, twitter:image, description, <title>).
- a local file (index.html, a JSX/TSX/MDX page, a built dist/) → read it directly.
- auto-detect from the codebase → search for the landing/marketing page (common spots: index.html, app/page.tsx, pages/index.*, src/App.*, landing/, marketing/, public/). Confirm the candidate with the user if ambiguous.
Codebase — a path to the repo (defaults to the current working directory). Used for the pricing/paywall/subscription principles, the feature surface ("does one thing"), and to locate the landing page if no URL/file was given.
Extra instructions (optional) — strategic context such as "we keep a free tier on purpose", "target audience is developers", "we must stay subscription". Honor these when interpreting a verdict (note the deliberate deviation) but still score the principle as written so the number stays comparable.

If neither a URL, a file, nor a detectable page exists, stop and ask the user where the landing page lives — do not invent one.

Pipeline (3 phases, in order)

Run these in sequence. Emit the matching Step Completion Report (see references/step-reports.md) after each.

Phase 1 — Resolve inputs & gather evidence

Resolve the landing page input (URL → /browse; file → read; else auto-detect).
Locate the codebase and find monetization evidence: billing SDKs (Stripe, Paddle, LemonSqueezy, RevenueCat, Chargebee), pricing config/constants, plan & tier definitions, paywall/auth gating, trial logic. Grep for price, plan, tier, checkout, subscription, free, trial, stripe, paddle.
Skim the feature surface (routes, nav items, top-level modules) to judge "does one thing".
For a large codebase, use grep/read (or a one-off Agent task scoped to pricing + feature evidence) so the main context stays clean. Collect: tier list, billing type (one-time vs subscription), free-plan yes/no, and a one-line feature inventory.
Note any extra instructions from the user.

Phase 2 — Evaluate against the 32 principles

Read references/principles.md — the full rubric. Score every principle PASS / PARTIAL / FAIL using its criteria. Do not skip any; absence of a thing a viral product would ship (pricing, testimonials, demo) is a real FAIL, not "unknown".
For each verdict, capture specific evidence from THIS product — quote the actual headline, name the actual tier, cite the file/line. Generic findings are not acceptable.
Tag every judgment/visual principle (hero punch, emotional headline, OG-image design, founder presence, novelty, price-vs-competitor) as low-confidence and record what a human must eyeball.
Compute the Virality Score: PASS=1, PARTIAL=0.5, FAIL=0, summed over 32, ×100/32, rounded. Assign the tier (Viral-ready / Promising / Needs work / Not viral yet).

Phase 3 — Prioritize fixes & write the report

Read references/report-template.md and produce the report in that exact shape: verdict block → scorecard (all 32) → top fixes (prioritized) → what's working → caveats.
The top fixes list is the core deliverable. Order by impact × ease, hero/paywall/ headline/proof/single-CTA first. Merge principles that share a root cause into one fix. Make each fix concrete enough to act on (give the actual proposed headline, the tier to cut, the CTA label) — quote what it is Now and what to Change it to.
Always write the report to viral-evaluation.md — repo root, or the current working directory when there is no repo — and also print the verdict block + top fixes inline.

Honest evaluation

This is a critique tool — its value is candor. Do not inflate scores to be encouraging. If the hero fails, say it fails and show the fix. At the same time, do not invent flaws: a genuine PASS is a PASS. Low-confidence verdicts must be labeled, never laundered into false certainty. When extra instructions justify a deliberate deviation (e.g. a strategic free tier), score the principle as written and explain the trade-off in the caveats — don't silently pass it.

On any failure to fetch inputs or read evidence, report the concrete error and stop — do not guess or continue with incomplete data. Confirm with the user on every gate and before finalizing the report. If a principle cannot be evidenced, mark FAIL or low-confidence; never invent.

Step Completion Reports

After each phase, emit the report from references/step-reports.md. The three phases are Gather Evidence, Evaluate, and Prioritize & Report.

Acceptance Criteria

All 32 principles scored with specific evidence quoted from the product.
Virality Score computed correctly (PASS=1, PARTIAL=0.5, FAIL=0) and tier assigned.
Top fixes are concrete, prioritized by impact×ease, with before/after suggestions.
Report always written to viral-evaluation.md (repo root, else the current working directory) and also printed inline; Step Completion Reports emitted per phase.
Negative-trigger domains respected (no SEO/ASO/copy/code-review work).

Expected output

A report containing:

Overall verdict + Virality Score (e.g. 68 — Promising)
Scorecard table for all 32 principles
Top 5-8 prioritized fixes with exact copy or code recommendations
What's already working
Caveats / low-confidence items

Edge cases

No landing page detectable: ask for URL or file; do not fabricate.
Codebase only (no public page): still score what can be inferred from code (pricing etc).
Strategic deviation justified by user (e.g. no testimonials by design): score as written, note the trade-off in caveats.
Partial evidence: mark affected principles low-confidence, never guess a PASS.

Reference files

references/principles.md — the 32-principle rubric: per-principle checks, evidence source, PASS/PARTIAL/FAIL bars, confidence flags, and the scoring formula. Load every run.
references/report-template.md — exact output shape (verdict block, scorecard, prioritized fixes, strengths, caveats) with a calibration example.
references/step-reports.md — Step Completion Report formats for the three phases.

Frequently asked questions

What does the Viral Product Evaluator AI skill do?

Review a product codebase and landing page against 32 viral principles and produce a Virality Score plus ranked fixes. Use to audit virality or prioritize growth. Don't use for SEO, ASO, copywriting, or code review.

Why use Viral Product Evaluator on TypingMind?

Because you install it once and use it with any model. Viral Product Evaluator is plain Markdown rather than provider-specific code, so the same skill runs on GPT-5, Claude, Gemini, Grok, or a local model — and you can switch model mid-chat without it breaking. TypingMind runs on your own API keys, so you pay providers directly instead of a per-seat subscription, and your skills and chats stay in your own storage.

How do I install Viral Product Evaluator in TypingMind?

Open Plugins → Skills → Install from GitHub in TypingMind and paste https://github.com/luongnv89/skills/tree/main/skills/viral-product-evaluator. TypingMind reads its SKILL.md and bundles its files and installs it as a skill you can enable per chat.

Which AI models can use Viral Product Evaluator?

Any model you connect in TypingMind. AI skills are plain Markdown instructions rather than provider-specific code, so GPT, Claude, Gemini, Grok, and local models can all load this skill when a request matches it.

How many AI models can I use with Viral Product Evaluator?

As many as you like. As long as a model supports skills, you can use Viral Product Evaluator with it — GPT, Claude, Gemini, Grok, DeepSeek, Mistral, Llama and more — all on TypingMind with your own API keys.

Is the Viral Product Evaluator AI skill free?

Yes. It is published on GitHub by luongnv89 under the MIT license. You only pay your own AI provider for the tokens you use.

What are AI skills?

An AI skill is a reusable instruction bundle that teaches an AI model how to do one specific task. It follows the open Agent Skills format: a SKILL.md file with a name and description, plus any scripts, templates or reference files the model may need. The model reads the instructions only when your request matches the skill, so an installed skill costs nothing until it is used.

How are AI skills different from plugins or MCP servers?

A plugin or MCP server gives a model new tools to call — code that runs somewhere and returns a result. An AI skill gives the model knowledge and process instead: how to approach a task, which steps to follow, what good output looks like. Skills are plain Markdown, so they need no server, no API key and no runtime, and they work with any model.

Overview

Installation

Install in TypingMind

Install in another agent

Use it in TypingMind

Works with any AI model

Skill instructions

Viral Product Evaluator

When to Use

Prerequisites

What this skill does and does not touch

Dependency Preflight (mandatory)

Inputs

Pipeline (3 phases, in order)

Phase 1 — Resolve inputs & gather evidence

Phase 2 — Evaluate against the 32 principles

Phase 3 — Prioritize fixes & write the report

Honest evaluation

Step Completion Reports

Acceptance Criteria

Expected output

Edge cases

Reference files

Bundled files

Frequently asked questions

Related AI Skills

Set up your own AI workspace now