We Audited Every GA4 Skill on ClawHub. Here's What the Source Code Revealed.

By BytesAgain | 2026-04-29

The 6,304-download champion isn't the only game in town. The 119-download unknown might be better.

---

The Setup

We ran a six-dimension source-code audit of 23 GA4-related skills on ClawHub. Not a description-skim. Not a download-count ranking. We installed every skill with working code, read every source file, compared architectures, and scored each one on code quality, feature coverage, authentication, output quality, extensibility, and community adoption.

Six had real code. Seven were documentation-only templates. Two were clones.

---

The Real Leaderboard

#	Skill	Score	Downloads	⭐	Type	Key Differentiator
1	ga4-analytics	89	6,304	11	TypeScript Platform	Only skill integrating GA4 + Search Console + Indexing API
2	google-analytics-ga4	88	119	0	Python Class	Only skill with realtime + historical + metadata modes
3	native-google-analytics	72	760	18	Pure REST	Zero SDK dependencies — calls the API directly with urllib
4	ga4	66	3,812	1	SDK Wrapper	Solid OAuth flow, but no realtime, no GSC
5	ga4-data-api	58	217	0	SDK Wrapper	Good install script, weakest query capabilities

Yes, the second-best GA4 skill on the platform has 119 downloads and zero stars.

---

The Hidden Gem

google-analytics-ga4 is the only GA4 skill on ClawHub that supports all three query modes:

• Realtime — --action realtime with configurable minute-range windows (0-30 minutes)
• Historical — --action historical with pagination, filtering, and offset
• Metadata — --action metadata listing every available dimension and metric

It outputs Markdown tables by default — no JSON parsing, no manual formatting. It has a separate openclaw_ga.py wrapper providing clean subcommand syntax. Its ga_query.py is a 280-line GoogleAnalyticsQuery class with proper error handling, response normalization, and totals aggregation.

Built by whosc. One version ever published. 119 downloads. Zero stars.

Why? The SKILL.md description buries the lead. It doesn't mention "realtime" anywhere prominent. The author built the most functionally complete GA4 tool on ClawHub — and forgot to market it.

Lesson #1: Code quality has almost zero correlation with downloads on ClawHub.

---

The Clone Problem

skill-ga4-analytics (390 downloads, zero2ai-hub) is an exact copy of ga4-analytics (adamkristopher).

We ran diff -rq on both source trees:

Files src/index.ts differ

The difference? One line in the reindexUrls() function:

- results.push({ url, status: 'submitted', ...result })
+ results.push({ ...result, url, status: 'submitted' })

405 lines. 10 files. 9 npm dependencies. Identical architecture, identical feature set, identical orchestration functions. Different author, different slug, different year — same code.

390 downloads for a copy-paste.

Lesson #2: The clone detection bar on ClawHub is extremely low.

---

Three Architectural Patterns

1. The TypeScript Platform (ga4-analytics)

scripts/src/
├── index.ts (405 lines) — 15 orchestration functions
├── api/reports.ts — 8 pre-built GA4 queries
├── api/realtime.ts — 4 realtime dimensions
├── api/searchConsole.ts — 6 GSC queries
├── api/indexing.ts — URL re-indexing
├── core/client.ts — Singleton: GA4+GSC+Indexing clients
├── core/storage.ts — Auto-save JSON results
└── config/settings.ts — .env driven

Heavy: 7 npm packages. Requires Node ≥18. But it's a complete data platform — site overview, traffic analysis, content performance, user behavior, date comparison, live snapshot, keyword analysis, SEO page performance, URL re-indexing, index status checking. All auto-saved with timestamps.

2. The Zero-Dependency REST (native-google-analytics)

# 175 lines. No pip install needed.
import urllib.request
url = f"https://analyticsdata.googleapis.com/v1beta/properties/{property_id}:runReport"
req = urllib.request.Request(url, data=payload, headers={"Authorization": f"Bearer {access_token}"})

The cleanest code in the category. 18 stars — highest of any GA4 skill. Developer love in its purest form: zero dependencies, calls the API directly, works on any Python 3.x installation.

3. The SDK Wrapper (ga4, ga4-data-api)

Standard google-analytics-data pip package with OAuth InstalledAppFlow. Functional but adds 50MB+ of virtual environment overhead. No realtime. No Search Console. No Indexing API. These are data extraction tools, not analysis platforms.

---

What Nobody Built Yet

Missing Feature	Current State	Opportunity
Chart/Graph Output	0/6 skills provide visualizations	First skill with matplotlib/plotly charts
Scheduled Reports	0/6 skills have cron integration	"Email me my GA4 report every Monday at 9am"
Notification Push	0/6 skills connect to messaging	Telegram/Slack/Discord webhook integration
Token Auto-Refresh	OAuth skills require manual re-auth	Service Account + automatic token refresh
Competitive Benchmarking	0/6 skills compare against norms	"Your bounce rate vs industry average"
Zero-Deps + Service Account	native-ga has zero deps but uses OAuth; ga4-analytics has Service Account but needs Node	Python + Service Account + urllib = ultimate combo

Lesson #3: The GA4 skill market is a land grab. First mover on any of these gaps wins.

---

The Overrated

ga4 has 3,812 downloads — second highest in the category — but scores only 66. It does one thing (historical report queries) competently. No realtime. No GSC. No Indexing. No auto-save. No orchestration.

Why the downloads? It has the best slug. ga4 is what everyone searches. Slug = distribution. Same lesson as domain names: the generic name gets the traffic regardless of quality.

---

Recommendations

If you're choosing a GA4 skill:

• Most complete: ga4-analytics — TypeScript, GA4+GSC+Indexing, auto-save, production-grade
• Hidden gem: google-analytics-ga4 — realtime+historical+metadata, Markdown output, 119 downloads is criminal
• Cleanest code: native-google-analytics — zero deps, pure REST, 18⭐ community approval
• Avoid: skill-ga4-analytics — it's a clone, use the original

If you're building something:

1. Don't compete on features — ga4-analytics already integrates GA4+GSC+Indexing. You can't beat that on breadth.
2. Compete on output — nobody has charts. Nobody has scheduled reports. Nobody has notifications.
3. Compete on simplicity — a single Python file with zero deps + Service Account auth beats every SDK wrapper.
4. Compete on the slug — if you can claim a generic, do it early.

---

Methodology: 6-dimension weighted scoring (Code Quality 20%, Feature Coverage 20%, Authentication 15%, Output Quality 15%, Extensibility 15%, Maintenance Heat 15%). All source code read and compared. Built with bytesagain-competitive-analysis.