Validate all skill packages before release

Browse files

Files changed (4) hide show

README.md +85 -489
docs/pipeline.md +47 -0
reports/2026-08-14-all-skills-execution-audit.md +50 -0
scripts/validate-skills.rb +99 -0

README.md CHANGED Viewed

@@ -1,527 +1,123 @@
----
-license: mit
-language:
-- en
-- zh
-- ja
-- ko
-pretty_name: "Gingiris Skills — Complete AI Growth Toolkit (42 Battle-Tested Playbooks for Claude Code, Cursor & OpenClaw)"
-task_categories:
-- text-generation
-- text-classification
-tags:
-- claude-code
-- claude-skill
-- agent-skill
-- ai-agent
-- ai-growth
-- growth-tools
-- growth-hacking
-- meta-router
-- skill-collection
-- llm-prompt
-- prompt-engineering
-- seo
-- geo
-- generative-engine-optimization
-- product-hunt
-- open-source
-- github-stars
-- b2b-saas
-- plg
-- aso
-- user-interview
-- kol-outreach
-- ugc-matrix
-- reddit-marketing
-- community-building
-- devrel
-- startup-marketing
-- saas-marketing
-- developer-tools
-- llm-tools
-- marketing
-- indie-hacker
-- ai-startup
-- AFFiNE
-- 30x-product-hunt
-size_categories:
-- n<1K
----
-# Gingiris Skills — Complete AI Growth Toolkit
-> **Give your AI agent 12 active execution skills backed by 42 battle-tested growth playbook datasets** — covers Product Hunt launches, GitHub stars, SEO/GEO, B2B SaaS, ASO, KOL outreach, Reddit marketing, and more. Use `/gr` to route each growth task to the narrowest installed specialist.
-## 📦 Install
 ```bash
 npx skills add Gingiris-1031/gingiris-skills
 ```
-Then ask your AI agent:
-> "I want to launch a SaaS on Product Hunt next month — full plan please" · "SEO traffic dropped 40% overnight, audit it" · "Help me design a Reddit campaign that won't get shadow-banned" · "Which playbook should I use for B2B PLG vs SLG?"
-🔗 [Browse the visual hub](https://gingiris.tools/skills/) · [Find growth tools](https://gingiris.tools/) · [Advisory services](https://gingiris.tools/services/) · [skills.sh listing](https://skills.sh/Gingiris-1031/gingiris-skills) · [ClawHub](https://clawhub.ai/gingiris-1031)
----
-## 📊 Citable Statistics
-| Stat | Value | Source / Context |
-|:---|:---|:---|
-| **Total playbooks bundled** | **42** | All Gingiris-* + thematic skills on HuggingFace |
-| **AFFiNE GitHub stars (organic)** | **60,000+** | 0→60K in 24 months (Aug 2022–Aug 2024) |
-| **Product Hunt #1 daily wins** | **30+** | Coached launches 2022–2026 |
-| **AI startup consultations** | **150+** | SEO/GEO/GTM advisory |
-| **gingiris.tools monthly impressions** | **~32,000** | March 2026, Google Search Console |
-| **Content publishing cadence** | **4 articles/week** | KD 30-50 long-tail focus |
-| **Multi-channel: HuggingFace + skills.sh + GitHub** | **3 distribution platforms** | Install via `npx skills add` |
-| **Language coverage** | **4** (EN / 中文 / 日本語 / 한국어) | All trigger keywords localized |
-| **Skill discovery via `/gr` meta-router** | **Single command** | Auto-routes to the matching specialist |
-**The thesis**: AI search engines (ChatGPT, Perplexity, Claude, Gemini) cite battle-tested playbooks with real numbers more than they cite generic SEO advice. Each Gingiris skill includes citable data points that improve both your agent's responses AND the long-term AI-search visibility of your product.
----
-## 🚀 30-Second Preview
-```
-You: /gr 我准备一个月后发 Product Hunt,需要完整规划
-Agent: ┌─ routing to gingiris-launch (Product Hunt specialist) ─┐
-       │ 4-week PH plan:                                          │
-       │   W-4: hunter outreach + asset gathering                 │
-       │   W-3: maker comments drafting + community warmup        │
-       │   W-2: launch day timeline + backup plans                │
-       │   W-1: dress rehearsal + KOL coordination                │
-       │ + auto-pulled: 30x PH #1 case study, hunter checklist   │
-       └──────────────────────────────────────────────────────────┘
-```
-One install gives the agent a router plus the active specialist workflows in this
-dataset. The GitHub monorepo contains the broader 55+ community-skill catalog.
----
-## 🛠️ The Toolbox (12 Active Skills + 42 Playbook Datasets)
-### Slash-Command Skills (v0.4.0)
-| Skill | Purpose |
-|:---|:---|
-| `/gr` | Meta-router — diagnoses your question, picks the matching specialist |
-| `/gr-seo-patrol` | Daily SEO/GEO patrol — SERP tracking, canonical fix, social-media avalanche rescue |
-| `/gr-blog-post` | Jekyll publishing — Iris voice + hreflang EN/CN/JA/KO + FAQ Schema |
-| `/gr-ph-launch` | Product Hunt launch playbook — 30x daily-#1 framework |
-| `/gr-oss-marketing` | Open-source go-to-market — GitHub stars + Reddit/HN/Discord distribution |
-| `/gr-b2b-growth` | B2B SaaS PLG/SLG, PMF to $10M ARR |
-| `/gr-aso` | App Store Optimization + mobile cold start |
-| `/gr-user-interview` | HeyGen 937-interview PMF methodology |
-| `/gr-competitor` | Competitor scanning via [actionbook](https://actionbook.dev) — 10x faster, 30-tab parallel |
-| `/gr-social-distill` | Blog → 4 social variants (X / 小红书 / LinkedIn / dev.to-Zenn) |
-| `/gr-geo-cite` | GEO citation tracking — weekly check across ChatGPT/Claude/Perplexity/Gemini |
-| `/gr-backlinks` | Systematic backlinks — Wikipedia / HARO-PR / G2 / Reddit-Quora 5 channels |
-### Roadmap (0.5+)
-| Skill | Source |
-|:---|:---|
-| `/gr-ph-comment` | Wraps [PH Comment Generator](https://gingiris.tools/tools/ph-comment-generator/) |
-| `/gr-gh-outreach` | Wraps [GitHub Issue Generator](https://gingiris.tools/tools/github-issue-generator/) |
-| `/gr-readme` | Wraps [GitHub README Generator](https://gingiris.tools/tools/github-readme-generator/) |
-| `/gr-hunter-radar` | actionbook-powered PH hunter activity scanner |
----
-## 🔄 The Workflow (how skills compose)
-```
-gr-competitor (see what competitors are doing)
-    ↓
-gr-ph-launch / gr-oss-marketing / gr-b2b (pick the play)
-    ↓
-gr-blog-post (create content)
-    ↓
-gr-seo-patrol (post-launch monitoring)
-    ↓ cannibalization        ↓ avalanche
-gr-seo-patrol canonical-fix  gr-seo-patrol rescue
-    ↓
-gr-user-interview (user feedback loop)
 ```
-Skills auto-recommend the next step:
-- `gr-ph-launch` 24h after publish → recommends `gr-seo-patrol` for monitoring
-- `gr-seo-patrol` detects cannibalization → auto-routes to canonical fix flow
-- `gr-blog-post` published → auto-adds article to `gr-seo-patrol` watchlist
 ---
-## ❓ FAQ
-**Q: What's the best Claude Code skill collection for AI/SaaS growth?**
-A: `gingiris-skills` bundles 42 battle-tested playbooks covering every growth dimension: Product Hunt launches (30+ #1 wins), GitHub stars (AFFiNE 0→60K case), SEO/GEO (32K monthly impressions), B2B SaaS PLG/SLG, ASO, KOL outreach, UGC matrix, Reddit marketing (40.11% LLM training share), user interviews (HeyGen 937 methodology), and competitor research. Install with `npx skills add Gingiris-1031/gingiris-skills` and use `/gr` as the meta-router.
-**Q: How is this different from generic "growth" Claude skills?**
-A: Every Gingiris skill is built from real campaigns, not theoretical advice. AFFiNE 60K stars, 30+ Product Hunt #1 daily wins, 150+ AI startup consultations, gingiris.tools 32K monthly impressions — these are the documented data points behind each playbook. Generic SEO skills give 2023-era advice (keyword density, backlinks); these include 2026 GEO patterns, JSON-LD templates that AI engines actually quote, and Reddit shadow-ban prevention.
-**Q: How do I install a single skill vs the whole bundle?**
-A: For the whole toolkit: `npx skills add Gingiris-1031/gingiris-skills`. For a single skill: `npx skills add Gingiris-1031/<slug>` — e.g. `npx skills add Gingiris-1031/gingiris-launch` for just Product Hunt. The complete index of 42 dataset slugs is in the "Full Playbook Index" section below.
-**Q: What does the `/gr` meta-router do?**
-A: `/gr` classifies direct tasks and broad growth problems, selects the narrowest installed specialist, runs it in the same task, and only hands off when the result creates a concrete next job. Ask "I'm launching on PH next month" and it routes to `/gr-ph-launch`; ask about a Reddit distribution problem and it routes to `/gr-oss-marketing`.
-**Q: Can I use these skills outside Claude Code?**
-A: Yes. They work in Cursor, OpenClaw, Codex CLI, Amp, Cline, and any agent that supports the SKILL.md standard. The HuggingFace dataset version is platform-agnostic — download the SKILL.md + references and use them as system prompts.
-**Q: Who built this?**
-A: [Iris Wei (生姜)](https://x.com/WeiYipei) — former cofounder/COO of [AFFiNE](https://github.com/toeverything/AFFiNE) ($10M raised, Forbes Asia 30 Under 30). Led AFFiNE 0→60K+ GitHub stars in 24 months. Now advises 150+ AI startups on SEO/GEO/GTM strategy.
 ---
-## 📚 Knowledge Base
-All methodology documents and atom-level knowledge points are open. Even without installing any skill, you can:
-### Structure
-```
-知识库/
-├── 原子库/
-│   ├── atoms.jsonl                    # Structured knowledge atoms (RAG-ready)
-│   └── README.md
-└── Skill知识包/
-    ├── iris_writing_style.md          # 5-element voice guide
-    └── seo_geo_playbook_2026.md       # SEO flywheel + GEO triple combo
-```
-### Usage Patterns
-**Pattern 1: Augment your AI's SEO capability**
-Paste `知识库/Skill知识包/seo_geo_playbook_2026.md` into your system prompt.
-**Pattern 2: Build a RAG**
-Load `atoms.jsonl` into your vector store. Each atom carries `topics` tags for filtering.
-**Pattern 3: Use a single script**
-`skills/gr-seo-patrol/scripts/*.py` runs standalone. See `docs/api-keys-template.md` for env config.
 ---
-## 🔧 Monthly Full-Site Audit Workflow
-> **Battle-tested 2026-05-07** on a 58-page Jekyll blog. Caught 43 SERP-truncating titles + 36 schema warnings + 27 stop-word slugs in a single 30-min run. One layout-level commit fixed 20 of 43 titles. **Use for any Jekyll / Hugo / Next.js blog with 30+ posts.**
-A repeatable 6-stage workflow you can run on any site. Powered by 4 scripts (attribution below).
-### Stage 1 — Discovery (5 min)
-Pull all blog URLs from your sitemap:
-```python
-import urllib.request, re
-sm = urllib.request.urlopen("https://your-site.com/sitemap.xml").read().decode()
-urls = [u for u in re.findall(r"<loc>([^<]+)</loc>", sm) if "/blog/" in u]
-```
-### Stage 2 — Parallel Audit (20 min for 60 pages)
-Run two audit scripts per URL in 4-thread parallel:
-```bash
-pip install requests
-python3 skills/gr-seo-patrol/scripts/check-page.py URL --timeout 20
-python3 skills/gr-seo-patrol/scripts/check-schema.py URL --timeout 20
-```
-Each script outputs a structured JSON envelope (`status: pass|warn|fail|info` per check).
-### Stage 3 — Aggregate Findings
-Bucket issues by type:
-- **Title length** > 70 chars (SERP truncation risk)
-- **H1 length** > 70 chars (mobile readability)
-- **Meta description** outside 80-170 chars
-- **Schema warns** by `@type` (BlogPosting / Article / Organization)
-- **Canonical mismatches**, **slug stop words**, **missing alt text**
-Save aggregated counts + per-URL lists to `findings.json`.
-### Stage 4 — Layered Fix Strategy (HIGH ROI ORDER)
-| Order | Layer | Scope | Typical commits | ROI |
-|---|---|---|---|---|
-| 1️⃣ | **Layout** (`_layouts/default.html`) | Schema bugs, title suffix, dateModified injection | 1 | 🔥 fixes 20+ pages at once |
-| 2️⃣ | **Config** (`_config.yml`) | Logo URL, twitter, social, author structure | 1 | fixes site-wide |
-| 3️⃣ | **Per-article batch** | Trim long titles/H1s, expand short meta | 10-20 | per-file, parallelizable |
-| 4️⃣ | **Skip** | Slug stop words (changing breaks 301), low-traffic old articles | 0 | low ROI |
-### Stage 5 — Verify
-After Jekyll/Hugo rebuild (~60-90s), re-run `check-schema.py` on a sample page. All schema types should show `status: pass`: Article · BlogPosting · Organization · FAQPage.
-### Stage 6 — Archive + Trend Track
-Commit `findings.json` to `data/audit-{YYYY-MM-DD}.json` for month-over-month trend analysis. Add 2-5 atoms to `知识库/原子库/atoms.jsonl` documenting any new lessons.
-### Schedule it
-```bash
-# In Claude Code's scheduled-tasks
-cronExpression: "0 10 1 * *"   # 10am on day 1 of each month
-prompt: "Run Monthly Full-Site Audit per gr-seo-patrol/SKILL.md workflow..."
-```
-### What you'll typically find on your first run
-Real numbers from gingiris.tools 2026-05-07 run:
-| Issue | Count | Resolution path |
-|---|---|---|
-| Title >70 chars | 43/58 | Layout-level (-20 chars suffix) + 13 per-article retrim |
-| Schema warns | 36 | Layout-level (dateModified + publisher.logo + contactPoint) |
-| H1 >70 chars | 23 | Per-article trim (paired with title) |
-| Meta too short/long | 20 | Per-article (i18n posts often hit this) |
-| Slug stop words | 27 | **SKIP** (would break 301 redirects) |
-| HTTP errors | 2 | Investigate (likely deleted/renamed) |
-**Total time**: ~30 min audit + 90 min fixes = **2 hours for site-wide SEO health refresh**.
-### HARD RULE (anti-hallucination guardrail)
-⛔ **Output ONLY the checks defined in the script's JSON envelope.**
-- Do NOT add "bonus" checks not in the script output
-- Do NOT contradict the script's `status` field without observable evidence
-- Do NOT invent metrics like "EEAT score 89" — third-party scoring is unofficial per Google 2026 guidance
-- If `llm_review_required: true`, make explicit judgment + document reasoning + update status
-The script envelope is **the single source of truth**. Treat as strict whitelist.
-### Script attribution
-The 4 audit scripts (`check-page.py`, `check-schema.py`, `check-site.py`, `check-social.py`) in `skills/gr-seo-patrol/scripts/` are **adapted from** [JeffLi1993/seo-audit-skill](https://github.com/JeffLi1993/seo-audit-skill) (MIT). Original repo focused on single-page client-presentable HTML reports; we adapted them for orchestrated batch audit + Jekyll/GitHub Pages site analysis. Original license terms preserved in each file header.
 ---
-## 🤝 About the Author
-**Iris Wei (生姜iris)** — Former cofounder & COO of [AFFiNE](https://github.com/toeverything/AFFiNE) ($10M raised, Forbes Asia 30 Under 30). Led AFFiNE from 0 to 60K+ GitHub stars across 100+ countries in 24 months.
-- Twitter: [@WeiYipei](https://x.com/WeiYipei)
-- Telegram: [@Iris_carrot](https://t.me/Iris_carrot)
-- Blog: [gingiris.tools](https://gingiris.tools)
-For 1-on-1 growth strategy review or advisory, reach via Telegram.
----
-## 🔗 Related Repositories
-- [Gingiris-1031/growth-tools](https://github.com/Gingiris-1031/growth-tools) — Main Jekyll blog, the live test bed for every skill ([gingiris.tools](https://gingiris.tools/blog/))
-- [dontbesilent2025/dbskill](https://github.com/dontbesilent2025/dbskill) — Framework inspiration
----
 ## License
-The downloadable community skills are licensed under MIT and remain free for
-personal, commercial, learning, and derivative use. Hosted Gingiris Pro
-execution, private datasets, managed services, and Gingiris brand assets are
-separate. See the repository's `TRADEMARKS.md` for brand-use rules.
-Use the free skills first. When a workflow needs live data or software, browse
-the curated [Gingiris growth tools directory](https://gingiris.tools/). For
-high-stakes decisions, ongoing execution support, or a customized AI growth
-employee, see [Gingiris advisory services](https://gingiris.tools/services/).
----
-## 🗂️ Full Playbook Index — 42 Skills Across 10 Categories
-The complete Gingiris playbook series on HuggingFace, organized by topic. Each dataset is installable via `npx skills add Gingiris-1031/<slug>` and queryable directly through your AI agent.
-### 🚀 Launch & Product Hunt (8)
-| Playbook | Focus |
-|:---|:---|
-| [gingiris-launch](https://huggingface.co/datasets/Gingiris/gingiris-launch) | Multi-channel launch sequencing, PH + KOL + UGC. 3 product paths: dev/maker (PH+GitHub), 2C consumer (communities/TikTok, skip PH), B2B (LinkedIn+media) |
-| [product-hunt-playbook](https://huggingface.co/datasets/Gingiris/product-hunt-playbook) | PH 30x #1 daily wins framework |
-| [product-hunt-launch-guide](https://huggingface.co/datasets/Gingiris/product-hunt-launch-guide) | T-14 to T+7 PH launch operations |
-| [ai-launch-playbook](https://huggingface.co/datasets/Gingiris/ai-launch-playbook) | AI product specific launch tactics |
-| [ai-product-launch](https://huggingface.co/datasets/Gingiris/ai-product-launch) | AI startup launch checklist |
-| [go-to-market-playbook](https://huggingface.co/datasets/Gingiris/go-to-market-playbook) | Complete 2026 GTM strategy |
-| [startup-launch](https://huggingface.co/datasets/Gingiris/startup-launch) | Startup launch fundamentals |
-| [startup-launch-playbook](https://huggingface.co/datasets/Gingiris/startup-launch-playbook) | Step-by-step startup launch SOP |
-### 🔍 SEO & GEO (2)
-| Playbook | Focus |
-|:---|:---|
-| [gingiris-seo-geo](https://huggingface.co/datasets/Gingiris/gingiris-seo-geo) | SEO + GEO dual-engine, AI search citation, 32K impressions case (Optimized for B2B/OSS. 2C products: see 2c-adaptation guide) |
-| [gingiris-seo-geo-agent](https://huggingface.co/datasets/Gingiris/gingiris-seo-geo-agent) | Autonomous SEO agent SOP, daily/weekly operations (Optimized for B2B/OSS. 2C products: see 2c-adaptation guide) |
-### 📈 B2B & SaaS (5)
-| Playbook | Focus |
-|:---|:---|
-| [gingiris-b2b-growth](https://huggingface.co/datasets/Gingiris/gingiris-b2b-growth) | B2B SaaS PLG/SLG, PMF to $10M ARR |
-| [saas-growth-playbook](https://huggingface.co/datasets/Gingiris/saas-growth-playbook) | SaaS scaling fundamentals |
-| [saas-marketing-playbook](https://huggingface.co/datasets/Gingiris/saas-marketing-playbook) | SaaS marketing channel mix |
-| [b2b-marketing-playbook](https://huggingface.co/datasets/Gingiris/b2b-marketing-playbook) | B2B campaign templates |
-| [plg-playbook](https://huggingface.co/datasets/Gingiris/plg-playbook) | Product-led growth motion design |
-### ⭐ Open Source (4)
-| Playbook | Focus |
-|:---|:---|
-| [gingiris-opensource](https://huggingface.co/datasets/Gingiris/gingiris-opensource) | OSS go-to-market, AFFiNE 0→60K stars |
-| [gingiris-github-star-growth](https://huggingface.co/datasets/Gingiris/gingiris-github-star-growth) | Monthly 300+ star sustained growth SOP |
-| [github-stars-playbook](https://huggingface.co/datasets/Gingiris/github-stars-playbook) | GitHub star tactical guide |
-| [open-source-marketing-playbook](https://huggingface.co/datasets/Gingiris/open-source-marketing-playbook) | OSS marketing channels & distribution |
-### 📱 Mobile & ASO (2)
-| Playbook | Focus |
-|:---|:---|
-| [gingiris-aso-growth](https://huggingface.co/datasets/Gingiris/gingiris-aso-growth) | ASO + app cold start + UGC creator matrix |
-| [aso-playbook](https://huggingface.co/datasets/Gingiris/aso-playbook) | App Store Optimization tactical guide |
-### 🤝 Community, KOL & Social (8)
-| Playbook | Focus |
-|:---|:---|
-| [gingiris-reddit-marketing](https://huggingface.co/datasets/Gingiris/gingiris-reddit-marketing) 🆕 | Reddit ops SOP — shadow ban prevention, AMA, 20-day Karma warming, **40.11% LLM training share** |
-| [gingiris-kol-outreach](https://huggingface.co/datasets/Gingiris/gingiris-kol-outreach) | KOL discovery to ROI tracking, AFFiNE 200+ campaigns |
-| [kol-outreach](https://huggingface.co/datasets/Gingiris/kol-outreach) | KOL cold outreach templates & DM scripts |
-| [gingiris-ugc-matrix](https://huggingface.co/datasets/Gingiris/gingiris-ugc-matrix) | UGC matrix scaling, Kuse $10M ARR / 60 days case |
-| [community-ambassador-playbook](https://huggingface.co/datasets/Gingiris/community-ambassador-playbook) | Ambassador program from recruitment to retention |
-| [viral-marketing-playbook](https://huggingface.co/datasets/Gingiris/viral-marketing-playbook) | Virality mechanics, network effects |
-| [devrel-playbook](https://huggingface.co/datasets/Gingiris/devrel-playbook) | DevRel: community, docs & events SOP |
-| [developer-marketing-playbook](https://huggingface.co/datasets/Gingiris/developer-marketing-playbook) | Developer-first marketing funnel |
-### 🎤 User Research (1)
-| Playbook | Focus |
-|:---|:---|
-| [gingiris-user-interview](https://huggingface.co/datasets/Gingiris/gingiris-user-interview) | User interview & PMF, HeyGen 937 methodology |
-### 🌱 Startup Growth & Strategy (8)
-| Playbook | Focus |
-|:---|:---|
-| [startup-growth-playbook](https://huggingface.co/datasets/Gingiris/startup-growth-playbook) | Early-stage startup growth fundamentals |
-| [startup-marketing-playbook](https://huggingface.co/datasets/Gingiris/startup-marketing-playbook) | Startup marketing channel selection |
-| [startup-consultant](https://huggingface.co/datasets/Gingiris/startup-consultant) | Strategic advisory framework |
-| [growth-hacking-playbook](https://huggingface.co/datasets/Gingiris/growth-hacking-playbook) | Experimentation & velocity tactics |
-| [growth-advisor](https://huggingface.co/datasets/Gingiris/growth-advisor) | Growth diagnostic framework |
-| [indie-hacker-playbook](https://huggingface.co/datasets/Gingiris/indie-hacker-playbook) | Solo founder / bootstrapped operations |
-| [competitor-research-playbook](https://huggingface.co/datasets/Gingiris/competitor-research-playbook) | Competitive intelligence + Lovable case study |
-| [product-dev-ops-playbook](https://huggingface.co/datasets/Gingiris/product-dev-ops-playbook) | Product & dev ops coordination SOP |
-### 🧭 AI Agent & Meta (3)
-| Playbook | Focus |
-|:---|:---|
-| [gingiris-growth-finder](https://huggingface.co/datasets/Gingiris/gingiris-growth-finder) | Meta-router: diagnoses situation, picks the right playbook. Includes Step 0 B2B vs B2C model selector |
-| [agent-workflow-playbook](https://huggingface.co/datasets/Gingiris/agent-workflow-playbook) | AI agent workflow design patterns |
-| [gingiris-go-global](https://huggingface.co/datasets/Gingiris/gingiris-go-global) | AI/SaaS overseas expansion full lifecycle (Phase 0-5) |
-### 📚 Hub & Blog (1)
-| Playbook | Focus |
-|:---|:---|
-| [growth-tools](https://huggingface.co/datasets/Gingiris/growth-tools) | Blog content + growth tools hub source |
----
-All 42 playbooks installable via `npx skills add Gingiris-1031/<slug>`. Browse the visual hub at [gingiris.tools/skills/](https://gingiris.tools/skills/) or list-form at [skills.sh/Gingiris-1031](https://skills.sh/Gingiris-1031).
-<!-- JSON-LD for SEO/GEO. AI engines (ChatGPT, Perplexity, Claude, Gemini) parse this. -->
-<script type="application/ld+json">
-{
-  "@context": "https://schema.org",
-  "@graph": [
-    {
-      "@type": "SoftwareApplication",
-      "name": "Gingiris Skills — Complete AI Growth Toolkit",
-      "applicationCategory": "DeveloperApplication",
-      "applicationSubCategory": "Claude Skill Collection",
-      "operatingSystem": "Any (Claude Code, Cursor, Codex, OpenClaw, Amp, Cline)",
-      "description": "Meta-router for 42 battle-tested growth playbooks. Covers Product Hunt launches (30x #1 wins), GitHub stars (AFFiNE 0→60K), SEO/GEO (32K monthly impressions), B2B SaaS PLG/SLG, ASO, Reddit marketing (40.11% LLM training share), KOL outreach, UGC matrix, user interviews (HeyGen 937 methodology), competitor research. Install via npx skills add Gingiris-1031/gingiris-skills.",
-      "url": "https://gingiris.tools/skills/",
-      "downloadUrl": "https://huggingface.co/datasets/Gingiris/gingiris-skills",
-      "installUrl": "https://skills.sh/Gingiris-1031/gingiris-skills",
-      "license": "https://opensource.org/licenses/MIT",
-      "offers": {
-        "@type": "Offer",
-        "price": "0",
-        "priceCurrency": "USD"
-      },
-      "author": {
-        "@type": "Person",
-        "name": "Iris Wei",
-        "alternateName": "生姜",
-        "url": "https://gingiris.tools",
-        "sameAs": [
-          "https://x.com/WeiYipei",
-          "https://www.linkedin.com/in/yipei-wei-550825105/"
-        ]
-      },
-      "keywords": "Claude Code skills, AI agent toolkit, growth playbooks, Product Hunt launch, GitHub stars, SEO GEO, B2B SaaS growth, ASO, KOL outreach, UGC matrix, Reddit marketing, user interview, competitor research, meta-router, gr command, AFFiNE 60K stars, 30x Product Hunt #1"
-    },
-    {
-      "@type": "FAQPage",
-      "mainEntity": [
-        {
-          "@type": "Question",
-          "name": "What's the best Claude Code skill collection for AI/SaaS growth?",
-          "acceptedAnswer": {
-            "@type": "Answer",
-            "text": "gingiris-skills bundles 42 battle-tested playbooks covering Product Hunt launches (30+ #1 wins), GitHub stars (AFFiNE 0→60K case), SEO/GEO (32K monthly impressions), B2B SaaS PLG/SLG, ASO, KOL outreach, UGC matrix, Reddit marketing (40.11% LLM training share), user interviews (HeyGen 937 methodology), and competitor research. Install with npx skills add Gingiris-1031/gingiris-skills and use /gr as the meta-router."
-          }
-        },
-        {
-          "@type": "Question",
-          "name": "How is this different from generic growth Claude skills?",
-          "acceptedAnswer": {
-            "@type": "Answer",
-            "text": "Every Gingiris skill is built from real campaigns, not theoretical advice. AFFiNE 60K stars, 30+ Product Hunt #1 daily wins, 150+ AI startup consultations, gingiris.tools 32K monthly impressions are the documented data points behind each playbook. Generic SEO skills give 2023-era advice; these include 2026 GEO patterns, JSON-LD templates that AI engines actually quote, and Reddit shadow-ban prevention."
-          }
-        },
-        {
-          "@type": "Question",
-          "name": "How do I install a single skill vs the whole bundle?",
-          "acceptedAnswer": {
-            "@type": "Answer",
-            "text": "For the whole toolkit: npx skills add Gingiris-1031/gingiris-skills. For a single skill: npx skills add Gingiris-1031/<slug>. The complete index of 42 dataset slugs is in the Full Playbook Index section."
-          }
-        },
-        {
-          "@type": "Question",
-          "name": "What does the /gr meta-router do?",
-          "acceptedAnswer": {
-            "@type": "Answer",
-            "text": "/gr classifies direct tasks and broad growth problems, selects the narrowest installed specialist, executes it in the same task, and only hands off when the result creates a concrete next job. Product Hunt requests route to /gr-ph-launch; Reddit distribution requests route to /gr-oss-marketing."
-          }
-        },
-        {
-          "@type": "Question",
-          "name": "Can I use these skills outside Claude Code?",
-          "acceptedAnswer": {
-            "@type": "Answer",
-            "text": "Yes. They work in Cursor, OpenClaw, Codex CLI, Amp, Cline, and any agent that supports the SKILL.md standard. The HuggingFace dataset version is platform-agnostic — download the SKILL.md + references and use them as system prompts."
-          }
-        },
-        {
-          "@type": "Question",
-          "name": "Who built this?",
-          "acceptedAnswer": {
-            "@type": "Answer",
-            "text": "Iris Wei (生姜) — former cofounder/COO of AFFiNE ($10M raised, Forbes Asia 30 Under 30). Led AFFiNE 0 to 60K+ GitHub stars in 24 months. Now advises 150+ AI startups on SEO/GEO/GTM strategy."
-          }
-        }
-      ]
-    }
-  ]
-}
-</script>

+# Gingiris Skills — AI Growth & Operations Toolkit
+> Reusable AI agent skills for startup growth operators. The downloadable
+> community skills are open source under the MIT License. Hosted Gingiris Pro
+> services, private datasets, and managed execution are separate commercial
+> products.
+> 4-language support: 🇺🇸 EN · 🇨🇳 ZH · 🇯🇵 JA · 🇰🇷 KR
+## ⚡ Install All 60 Skills in One Command
 ```bash
 npx skills add Gingiris-1031/gingiris-skills
 ```
+Or install a single skill:
+```bash
+npx skills add Gingiris-1031/gr-readme
+npx skills add Gingiris-1031/gr-oss-marketing
+# ... any skill from the index below
 ```
+Browse all skills: [gingiris.tools/skills](https://gingiris.tools/skills/) · [clawhub.ai/gingiris-1031](https://clawhub.ai/gingiris-1031)
 ---
+## 🇨🇳 中文简介
+这是 Gingiris 增长技能 monorepo，收录 60 个可复用的 Claude Code skills，覆盖 SEO/GEO、产品发布、开源营销、B2B 增长、用户访谈等出海增长场景。所有内容提炼自真实操盘经验，而非二手资料汇编。一条命令安装全部：`npx skills add Gingiris-1031/gingiris-skills`，也可以按上面的索引单独安装某一个 skill。完整技能列表与使用说明见 [gingiris.tools/skills](https://gingiris.tools/skills/)。
+## 🇯🇵 日本語
+Gingiris 成長スキルのモノレポです。SEO/GEO、プロダクトローンチ、オープンソースマーケティング、B2B グロース、ユーザーインタビューなどをカバーする 60 個の Claude Code スキルを収録しています。`npx skills add Gingiris-1031/gingiris-skills` の 1 コマンドで一括インストールでき、個別スキルのみのインストールにも対応しています。詳しいスキル一覧は [gingiris.tools/skills](https://gingiris.tools/skills/) をご覧ください。
+## 🇰🇷 한국어
+Gingiris 성장 스킬 모노레포입니다. SEO/GEO, 제품 출시, 오픈소스 마케팅, B2B 성장, 사용자 인터뷰 등 글로벌 그로스 시나리오를 다루는 60 개의 Claude Code 스킬을 담고 있습니다. `npx skills add Gingiris-1031/gingiris-skills` 명령 하나로 전체를 설치할 수 있고, 필요한 스킬만 골라 설치할 수도 있습니다. 전체 스킬 목록과 사용법은 [gingiris.tools/skills](https://gingiris.tools/skills/)에서 확인하세요.
 ---
+## Skills Index
+### Core Growth Playbooks (`gingiris-*`)
+| Skill | Description | ClawHub |
+|-------|-------------|---------|
+| [gingiris-b2b-growth](skills/gingiris-b2b-growth) | B2B SaaS Growth — PLG/SLG Playbook, $0→$10M ARR | [↗](https://clawhub.ai/gingiris-1031/skills/gingiris-b2b-growth) |
+| [gingiris-aso-growth](skills/gingiris-aso-growth) | ASO & App Cold Start — Organic Discovery & UGC Growth | [↗](https://clawhub.ai/gingiris-1031/skills/gingiris-aso-growth) |
+| [gingiris-seo-geo](skills/gingiris-seo-geo) | SEO & GEO 2026 — Rank Google + AI Search | [↗](https://clawhub.ai/gingiris-1031/skills/gingiris-seo-geo) |
+| [gingiris-seo-geo-agent](skills/gingiris-seo-geo-agent) | SEO/GEO Agent Ops — 32K Impressions Monthly Autopilot | [↗](https://clawhub.ai/gingiris-1031/skills/gingiris-seo-geo-agent) |
+| [gingiris-go-global](skills/gingiris-go-global) | AI Product Go-Global — Research to Monetization SOP + **Hardware Market Entry Strategy (Plaud Japan Makuake, Insta360 Regional Office Model, 15% GMV trigger)** | [↗](https://clawhub.ai/gingiris-1031/skills/gingiris-go-global) |
+| [gingiris-launch](skills/gingiris-launch) | Product Launch — Multi-Channel GTM & Cold Start | [↗](https://clawhub.ai/gingiris-1031/skills/gingiris-launch) |
+| [gingiris-opensource](skills/gingiris-opensource) | Open Source Marketing — GitHub Stars Growth System | [↗](https://clawhub.ai/gingiris-1031/skills/gingiris-opensource) |
+| [gingiris-github-star-growth](skills/gingiris-github-star-growth) | GitHub Star Sustained Growth — 300+ Stars/Month | [↗](https://clawhub.ai/gingiris-1031/skills/gingiris-github-star-growth) |
+| [gingiris-kol-outreach](skills/gingiris-kol-outreach) | KOL Outreach & Influencer Marketing — Discovery to ROI + **Hardware KOL Vertical Matrix (Plaud/Insta360 cases)** | [↗](https://clawhub.ai/gingiris-1031/skills/gingiris-kol-outreach) |
+| [gingiris-ugc-matrix](skills/gingiris-ugc-matrix) | UGC Content Matrix — Viral Distribution System + **Hardware UGC Flywheel (Insta360 NoseMode, DJI SkyPixel vs Algorithm)** | [↗](https://clawhub.ai/gingiris-1031/skills/gingiris-ugc-matrix) |
+| [gingiris-growth-finder](skills/gingiris-growth-finder) | Growth Finder — AI Strategy Router for Startups | [↗](https://clawhub.ai/gingiris-1031/skills/gingiris-growth-finder) |
+| [gingiris-user-interview](skills/gingiris-user-interview) | User Interview & PMF Validation — JTBD Framework (937 Interviews) | [↗](https://clawhub.ai/gingiris-1031/skills/gingiris-user-interview) |
+| [community-ambassador-playbook](skills/community-ambassador-playbook) | Community Ambassador Playbook — Recruitment to Retention SOP + **Hardware Brand Cases (Insta360 4-tier creator system, DJI 3-track model, Official Takeover mechanism)** | [↗](https://clawhub.ai/gingiris-1031/skills/community-ambassador-playbook) |
+| [competitor-research-playbook](skills/competitor-research-playbook) | Competitor Research Playbook — CI to GTM | [↗](https://clawhub.ai/gingiris-1031/skills/competitor-research-playbook) |
+| [product-hunt-launch-playbook-win-1-daily](skills/product-hunt-launch-playbook-win-1-daily) | Product Hunt Launch Playbook — Win #1 Daily | [↗](https://clawhub.ai/gingiris-1031/skills/product-hunt-launch-playbook-win-1-daily) |
+| [hardware-saas-gtm](skills/hardware-saas-gtm) | Hardware + SaaS GTM Playbook — Crowdfunding Engineering, KOL Matrix & Device-as-Funnel Subscription (Plaud/Insta360/DJI cases) | [↗](https://clawhub.ai/gingiris-1031/skills/hardware-saas-gtm) |
+### Execution Skills (`gr-*`)
+| Skill | Description |
+|-------|-------------|
+| [gr](skills/gr) | Growth Router — Entry point for all gr-* skills |
+| [gr-aso](skills/gr-aso) | ASO Execution — App Store keyword & rating optimization |
+| [gr-b2b-growth](skills/gr-b2b-growth) | B2B Growth Execution — PLG funnel & enterprise motion |
+| [gr-backlinks](skills/gr-backlinks) | Backlinks & Link Building — 5-channel SEO authority system |
+| [gr-blog-post](skills/gr-blog-post) | Blog & Content Marketing — SEO-optimized publishing pipeline |
+| [gr-competitor](skills/gr-competitor) | Competitor Analysis Execution — CI workflow |
+| [gr-geo-cite](skills/gr-geo-cite) | GEO & AI Search Citations — Get cited by ChatGPT/Perplexity |
+| [gr-oss-marketing](skills/gr-oss-marketing) | Open Source Marketing Execution — GitHub ecosystem growth |
+| [gr-ph-launch](skills/gr-ph-launch) | Product Hunt Launch Execution — T-14 to T+7 playbook |
+| [gr-seo-patrol](skills/gr-seo-patrol) | SEO Audit & Monitoring — Technical SEO weekly patrol |
+| [gr-social-distill](skills/gr-social-distill) | Social Media Distillation — Content repurposing system |
+| [gr-user-interview](skills/gr-user-interview) | User Interview Execution — PMF validation workflow |
+| [gr-readme](skills/gr-readme) | GitHub README Writing System — 3-Second Conversion Framework (AFFiNE 0→60K Stars) |
 ---
+## Tools Used in These Playbooks
+| Tool | What It Does | Used In |
+|------|-------------|--------|
+| [Analook](https://www.analook.com) | Competitive intelligence — URL → SEO/traffic/social report in 5 min | gr-competitor, gingiris-growth-finder, competitor-research-playbook |
+| [Analook Growth Audit](https://www.analook.com/growth-audit.html) | Deep growth audit for any product ($29/mo) | gingiris-b2b-growth, gingiris-launch |
+> Analook is built by Iris Wei. Use code **GINGIRIS20** for 20% off.
 ---
+## Author
+**Iris Wei** · [GitHub](https://github.com/Gingiris-1031) · [Twitter @WeiYipei](https://x.com/WeiYipei) · [gingiris.tools](https://gingiris.tools)
+Built from real operator experience across 14 countries, 34 ambassador programs, and 937 user interviews.
 ## License
+Unless a subdirectory states otherwise, the source code, prompts, templates, and
+skill files in this repository are available under the [MIT License](LICENSE).
+You may use, modify, redistribute, and commercially use these repository
+materials under its terms.
+The MIT License applies only to materials actually distributed in this
+repository. It does not include:
+- hosted Gingiris Pro APIs or managed skill execution;
+- private datasets, credentials, infrastructure, or server-side workflows;
+- paid reports, advisory deliverables, service levels, or customer data;
+- the Gingiris names, logos, domains, or other brand assets.
+Use of Gingiris branding is governed separately by the
+[Gingiris Brand and Trademark Policy](TRADEMARKS.md). The license does not imply
+endorsement or permission to present a fork, service, or product as official
+Gingiris.
+Commercial licensing for future self-hosted Pro components, white-label use, or
+OEM distribution may be offered separately. Contact Gingiris through
+[gingiris.tools](https://gingiris.tools/).

docs/pipeline.md ADDED Viewed

	@@ -0,0 +1,47 @@

+# Skills 三层结构与同步管线
+> 最后更新：2026-07-06。monorepo 是 ClawHub / skills.sh 的分发源和 SEO 资产：
+> **不删除 skill 目录、不缩减对外 description、不减少对外内容。**
+## 三层结构
+| 层 | 位置 | 身份 | 能否手改 |
+|---|---|---|---|
+| 1. 分发源 | `skills/<name>/` | 唯一可编辑源，全量内容（含完整多语 description） | ✅ 只改这里 |
+| 2. 嵌套打包产物 | `skills/<name>/<name>/` | `scripts/sync-nested.sh` 生成的打包副本（npx skills add 只装 SKILL.md 所在目录，嵌套层保住 references/） | ❌ 禁止手改 |
+| 3. 本机运行时 | `~/.claude/skills/<name>/`（多为 symlink → `~/.agents/skills/<name>/`） | `scripts/sync-local.py` 生成的精简版构建产物（description 截 ~320 字符 + 前 8 个 Triggers，省 context） | ❌ 禁止手改 |
+## 改动流程
+```
+改 skills/<name>/ 下的源文件
+  → bash scripts/sync-nested.sh    # 重建嵌套打包副本（字节级一致校验）
+  → python3 scripts/sync-local.py  # 重建本机精简副本（幂等）
+  → ruby scripts/validate-skills.rb # 全量结构、引用、隐私与打包校验
+完成
+```
+## 防线与规则
+- **`scripts/check-dead-refs.sh` 是 pre-commit 防线**：commit 前必须过
+  （拦 gingiris.github.io / 老账号 github.com/Gingiris / gingiris.com；
+  `*@gingiris.com` 邮箱豁免）。
+- 私有文件不进分发：`sync-nested.sh` 顶部的 `EXCLUDES` 数组
+  （现有 gingiris-twitter-agent-ops 的 `references/weiyipei-ops.md`、
+  gr-backlinks 的 `data/`），新增私有文件直接加条目。
+- 本机孤儿 skill（monorepo 没有的，如 gingiris-reddit-marketing）：
+  sync-local.py 不覆盖、只精简 description（原文备份为 `SKILL.md.orig`）；
+  源码防丢备份在本仓 `archive/`（不在 skills/ 内，不进分发）。
+## 全量执行验收
+每次发布不抽查旗舰 skill，而是对 `skills/*/SKILL.md` 全量验收。发布数以实际顶层目录为准，不手填在多个平台。发布前依次运行：
+```bash
+bash scripts/sync-nested.sh
+ruby scripts/validate-skills.rb
+bash scripts/check-dead-refs.sh
+GINGIRIS_SKILL_REPORT=/tmp/gingiris-skill-report.json ruby scripts/validate-skills.rb
+```
+`validate-skills.rb` 会逐个解析 frontmatter、核对目录名与 `name`、检查嵌套安装包字节一致性、阻止私有文件进入公开包，并扫描旧 GitHub 身份、旧站点和错误 ClawHub handle。非零退出码即禁止发布到 GitHub、Hugging Face、ClawHub、gingiris.tools 或 skills.sh。

reports/2026-08-14-all-skills-execution-audit.md ADDED Viewed

	@@ -0,0 +1,50 @@

+# All Skills Execution Audit — 2026-08-14
+## Scope
+- Canonical source: `Gingiris-1031/gingiris-skills`
+- Skills executed through the release gate: **60/60**
+- Flagship skills represented on gingiris.tools: launch, OSS/GitHub, SEO/GEO,
+  B2B/PLG, ASO, KOL/UGC, community, competitor research, user interviews,
+  go-global, Twitter operations, growth routing, and hardware + SaaS GTM
+- Distribution targets: GitHub, Hugging Face, ClawHub, gingiris.tools, skills.sh
+## Result
+| Check | Result |
+|---|---:|
+| YAML frontmatter parsed | 60/60 |
+| Directory name matches skill `name` | 60/60 |
+| Required `description` present | 60/60 |
+| Nested packages byte-identical | 12/12 |
+| Private files excluded from public packages | PASS |
+| Legacy GitHub identity / dead public references | 0 |
+| Blocking errors | 0 |
+| Trigger-language improvement warnings | 36 |
+The 36 warnings are non-blocking: those skills already have valid descriptions
+and install correctly, but their discovery text does not use explicit phrases
+such as “use this skill when” or “triggers”. They should be improved in normal
+content batches without inflating every agent's startup context at once.
+## Context-specific improvements
+1. The public count is now tied to the real top-level inventory: **60**, replacing
+   the inconsistent “40+” website copy and “55+” repository copy.
+2. The validator encodes Gingiris account safety: only `Gingiris-1031` is valid;
+   legacy GitHub, Pages, ClawHub handle, and gingiris.com references block release.
+3. Twitter private operations notes and backlink internal ledgers are checked for
+   leakage before any public package is accepted.
+4. The release gate emits a machine-readable JSON report so hosted skill runs and
+   future CI can use the same source of truth.
+5. Distribution remains source-first: edit the monorepo, rebuild nested/local
+   artifacts, validate once, then publish mirrors. skills.sh indexes the GitHub
+   source instead of becoming a separately edited copy.
+## Re-run
+```bash
+bash scripts/sync-nested.sh
+GINGIRIS_SKILL_REPORT=/tmp/gingiris-skill-report.json ruby scripts/validate-skills.rb
+bash scripts/check-dead-refs.sh
+```

scripts/validate-skills.rb ADDED Viewed

	@@ -0,0 +1,99 @@

+#!/usr/bin/env ruby
+# frozen_string_literal: true
+# Zero-dependency release gate for every distributable Gingiris skill.
+require "json"
+require "yaml"
+require "pathname"
+ROOT = Pathname.new(__dir__).parent
+SKILLS = ROOT.join("skills")
+BANNED = [
+  %r{github\.com/Gingiris/}, %r{gingiris\.github\.io},
+  %r{clawhub\.ai/(?:user/)?gingiris/}i,
+  %r{https?://gingiris\.com(?:/|\b)}
+].freeze
+PRIVATE_PATHS = {
+  "gingiris-twitter-agent-ops" => ["references/weiyipei-ops.md"],
+  "gr-backlinks" => ["data"]
+}.freeze
+errors = []
+warnings = []
+rows = []
+SKILLS.children.select(&:directory?).sort.each do |dir|
+  slug = dir.basename.to_s
+  skill_file = dir.join("SKILL.md")
+  unless skill_file.file?
+    errors << "#{slug}: missing SKILL.md"
+    next
+  end
+  text = skill_file.read
+  match = text.match(/\A---\n(.*?)\n---\n/m)
+  unless match
+    errors << "#{slug}: invalid or missing YAML frontmatter"
+    next
+  end
+  begin
+    meta = YAML.safe_load(match[1], permitted_classes: [], aliases: false)
+  rescue StandardError => e
+    errors << "#{slug}: YAML parse failed: #{e.message}"
+    next
+  end
+  name = meta.is_a?(Hash) ? meta["name"].to_s : ""
+  description = meta.is_a?(Hash) ? meta["description"].to_s.strip : ""
+  errors << "#{slug}: name is #{name.inspect}, expected #{slug.inspect}" unless name == slug
+  errors << "#{slug}: description is empty" if description.empty?
+  warnings << "#{slug}: description has no explicit trigger language" unless description.match?(/trigger|use this skill|when you need|适用|触发/i)
+  public_files = dir.glob("**/*").select(&:file?).reject do |path|
+    relative = path.relative_path_from(dir).to_s
+    relative.start_with?("#{slug}/") || PRIVATE_PATHS.fetch(slug, []).any? do |private_path|
+      relative == private_path || relative.start_with?("#{private_path}/")
+    end
+  end
+  public_files.each do |path|
+    next unless %w[.md .json .yaml .yml .txt .html].include?(path.extname)
+    body = path.read
+    BANNED.each do |pattern|
+      errors << "#{path.relative_path_from(ROOT)}: banned public reference #{pattern.inspect}" if body.match?(pattern)
+    end
+  end
+  nested = dir.join(slug)
+  if nested.directory?
+    public_files.each do |source|
+      relative = source.relative_path_from(dir)
+      target = nested.join(relative)
+      errors << "#{slug}: nested package missing #{relative}" unless target.file?
+      errors << "#{slug}: nested package differs at #{relative}" if target.file? && source.binread != target.binread
+    end
+    PRIVATE_PATHS.fetch(slug, []).each do |private_path|
+      errors << "#{slug}: private path leaked into nested package: #{private_path}" if nested.join(private_path).exist?
+    end
+  end
+  body = text.sub(match[0], "")
+  rows << {
+    slug: slug, bytes: text.bytesize,
+    references: public_files.count { |path| path.to_s.include?("/references/") },
+    scripts: public_files.count { |path| path.to_s.include?("/scripts/") },
+    nested: nested.directory?,
+    execution_sections: body.scan(/^##+\s+.*(?:执行|execution|workflow|SOP|output|输出|步骤|step)/i).length
+  }
+end
+expected = SKILLS.children.count(&:directory?)
+errors << "inventory mismatch: found #{rows.length} valid skills out of #{expected}" unless rows.length == expected
+summary = {
+  generated_at: Time.now.utc.strftime("%Y-%m-%dT%H:%M:%SZ"), skills_checked: expected,
+  passed: errors.empty?, error_count: errors.length, warning_count: warnings.length,
+  with_nested_packages: rows.count { |row| row[:nested] },
+  with_scripts: rows.count { |row| row[:scripts].positive? }, skills: rows
+}
+puts JSON.pretty_generate(summary.reject { |key, _| key == :skills })
+warnings.each { |warning| warn "WARN: #{warning}" }
+errors.each { |error| warn "ERROR: #{error}" }
+report_path = ENV["GINGIRIS_SKILL_REPORT"]
+File.write(report_path, JSON.pretty_generate(summary) + "\n") if report_path && !report_path.empty?
+exit(errors.empty? ? 0 : 1)