DEV Community: TeHsuWang

I Built This Tool With Three AIs at Once — Claude, Gemini, and Copilot

TeHsuWang — Fri, 10 Apr 2026 08:12:30 +0000

Most AI-assisted development follows a simple pattern: open a chat, describe the problem, copy the output, adjust, repeat. One AI, one conversation, one step at a time.

aeoptimize was built differently.

It's an open-source CLI that scores websites for AI readability — how likely your content is to be cited by ChatGPT, Perplexity, or Google AI Overview. During development, we dispatched different components to different AI systems in parallel. The tool does the same thing at runtime.

The division of labor

We split the work by capability fit, not convenience:

Claude: core architecture, scoring engine, security audit
Gemini: Vite plugin
Copilot: Next.js plugin

Claude handles long-context reasoning well, which suited the 17-rule scoring engine and the adversarial review. Gemini and Copilot each received a self-contained plugin with a clear interface, where code generation speed mattered more than sustained context. The two plugins also needed to conform to their respective framework conventions, and each AI had stronger exposure to one ecosystem's idioms.

What each AI produced

Claude built the core: the HTML/Markdown parser, 17 deterministic scoring rules across 5 dimensions, the llms.txt generator, JSON-LD schema generator, and the pre-commit hook system. Four rounds of security review — SSRF protection, shell injection prevention, staged content scanning — were also Claude's work.

Gemini built the Vite plugin (~52 lines). Hooks into configResolved and closeBundle to auto-generate llms.txt and JSON-LD at build time. Clean on the first pass.

Copilot built the Next.js plugin (~55 lines). Same behavior, different framework conventions. Also clean.

Both ship under the same npm package with separate export paths (aeoptimize/vite, aeoptimize/next).

The adversarial review

After the first stable version, we ran a Codex security review — treating it as an attacker, not a collaborator.

Four HIGH severity findings:

GitHub Action using unpinned npx aeoptimize@latest — any future compromised version runs automatically in CI
Pre-commit hook scoring the working tree instead of the staged diff
Hook install/uninstall replacing the entire .git/hooks/pre-commit, destroying other hooks in the process
SSRF bypass via redirects — following hops without validating each destination, which could resolve a public URL to an internal IP

All four fixed in v0.5.2. The GitHub Action now uses bundled $GITHUB_ACTION_PATH/../dist/cli/index.js. The hook reads staged content via git show :"$FILE". Install/uninstall uses # BEGIN/END aeoptimize delimited blocks. Each redirect hop is now validated against private IP ranges before following.

The redirect issue is worth dwelling on. The original code validated the initial URL — that seemed sufficient. An adversarial reviewer asked what happened when a validated URL redirected somewhere else. The answer was not good. That kind of question is harder to ask about code you wrote.

Multi-AI scoring at runtime

The same pattern extends to the tool itself:

npx aeoptimize scan your-site.com --multi-ai

With --multi-ai, the tool checks whether gemini and copilot CLIs are installed, sends each the page content for independent evaluation, then merges results with the rule engine:

2 AIs available: 50/50 between rule score and AI consensus
1 AI available: 60/40 in favor of the rule engine
No AIs: 100% rule engine

The rule engine runs offline and is fully deterministic — the base scan has no cost and no rate limits. The AI layer adds qualitative insight when available.

Score: 72/100 (Rule Engine: 61 | AI Consensus: 83)

AI Insights:
  Gemini:  "Missing llms.txt reduces discoverability by AI crawlers"
  Copilot: "FAQ section present but lacks FAQPage schema markup"

Why this works

The redirect vulnerability explains it well enough. The developer who wrote the code also validated it. The validation made sense given what the developer intended. An external reviewer — one not trying to make the code work — looked at the same function and found the gap.

Independent evaluation surfaces assumptions the original author couldn't see, because those assumptions were load-bearing when the code was written.

That applies to security review. It also applies to content quality, schema completeness, and whatever else you're trying to score consistently.

Try it

npx aeoptimize scan your-site.com
npx aeoptimize scan your-site.com --multi-ai

GitHub: https://github.com/dexuwang627-cloud/aeoptimize
npm: https://www.npmjs.com/package/aeoptimize

If you've used a multi-AI workflow in your own work, curious how you split it up.

Is Your Website Invisible to AI? Stop Optimizing for Google, Start Optimizing for ChatGPT

TeHsuWang — Wed, 08 Apr 2026 08:27:58 +0000

67% of users now get their first answer from an AI assistant rather than clicking a search result. Your site might rank #1 on Google and still be invisible to ChatGPT, Perplexity, and Google AI Overview.

The problem: SEO optimizes for ranking algorithms. AI engines don't rank -- they cite. If your content can't be extracted and quoted, it doesn't exist.

This is AEO (Answer Engine Optimization), and we built an open-source tool for it.

One Command

npx aeoptimize scan your-site.com

AEO Readability Report
Score: 66/100  AI Readability: Good

  Structure        ██████████████░░░░░░ 18/25
  Citability       █████████████░░░░░░░ 16/25
  Schema           ███████░░░░░░░░░░░░░  7/20
  AI Metadata      █████████████████░░░ 13/15
  Content Density  ████████████████░░░░ 12/15

Warnings:
  ! Found 2 H1 headings. Use exactly one H1 per page.
  ! JSON-LD missing fields: @type, name, description

Top Suggestions:
  -> Add FAQ section with question-format headings
  -> Add AI-relevant schema types (FAQPage, Article, HowTo)

No API keys. No signup. Runs offline.

What aeoptimize Measures

17 rules across 5 dimensions, scored 0-100:

Dimension	Max	What AI engines care about
Structure	25	Clear headings, short paragraphs (<150 words), FAQ sections
Citability	25	Self-contained statements an AI can quote without context
Schema	20	JSON-LD that LLMs use as source of truth (FAQPage, Article)
AI Metadata	15	llms.txt file, robots.txt AI crawler rules
Content Density	15	Signal-to-noise ratio, vocabulary diversity

The scoring is deterministic -- same input, same score. No LLM involved in the base scan, so you can run it in CI/CD without worrying about cost or rate limits.

Benchmarking Real Websites

Website	Score	Why
nextjs.org/docs	75	Good structure, has llms.txt, missing some schema
shiheintelligent.com	66	Has schema but incomplete, 2 H1s, no FAQ
stripe.com/docs	59	Great content but zero JSON-LD (-20 points)
anthropic.com	52	Landing page with little extractable content

Even well-built sites lose 20+ points from missing JSON-LD alone. That's the difference between being cited and being ignored.

Case Study: Fixing a 66-Score Site

shiheintelligent.com scored 66/100. The three biggest issues:

Two H1 headings -- AI engines treat H1 as the page topic. Two H1s create ambiguity.
Incomplete JSON-LD -- Schema existed but missing @type, name, description. AI crawlers can't categorize the content.
No FAQ section -- The site answers common questions in paragraph form, but without FAQ schema, AI engines can't extract them as Q&A pairs.

Fixing these three issues alone could push the score above 80.

More Than a CLI

aeoptimize works in multiple modes:

Framework plugins -- auto-generate llms.txt and JSON-LD on build:

// vite.config.ts
import { aeoPlugin } from 'aeoptimize/vite';
export default defineConfig({ plugins: [aeoPlugin()] });

// next.config.mjs
import { withAeo } from 'aeoptimize/next';
export default withAeo({});

Pre-commit hook -- block commits that drop AEO score:

npx aeoptimize hook install              # min score: 60
npx aeoptimize hook install --min-score 80

GitHub Action -- check every PR:

- uses: dexuwang627-cloud/aeoptimize/action@main
  with:
    path: dist
    min-score: 60

Claude Code skills -- interactive, AI-powered optimization:

claude plugin marketplace add dexuwang627-cloud/aeoptimize
# Then: /aeo-scan, /aeo-generate, /aeo-transform

The /aeo-transform skill uses your Claude subscription to restructure content: split paragraphs, extract FAQ schemas, remove keyword stuffing, fix dangling references. Zero extra cost.

Multi-AI Scoring

For higher confidence, aeoptimize can dispatch scoring to multiple AI engines simultaneously:

npx aeoptimize scan your-site.com --multi-ai

It detects gemini and copilot CLIs on your system, sends each the page content for independent evaluation, then merges with the rule engine using weighted consensus. The result shows per-AI insights:

Score: 72/100 (Rule Engine: 61 | AI Consensus: 83)

AI Insights:
  Claude:  "FAQ section lacks schema markup"
  Gemini:  "Missing llms.txt reduces discoverability"

Built by Three AIs

The project itself was built as a multi-AI collaboration:

Claude -- architecture, core engine (17 rules, scanner, generator), security audit
Gemini -- Vite plugin
Copilot -- Next.js plugin

50 tests, 4 rounds of automated code review (code quality, security, architecture, final). The biggest competitor in this space has 4,800+ stars but zero tests.

The SEO vs AEO Shift

	SEO	AEO
Goal	Rank higher	Get cited
Audience	Search crawler	Language model
Key metric	Click position	Citation accuracy
Content style	Keyword-rich	Self-contained, structured
Structured data	Nice to have	Essential

This shift is already happening. Google AI Overview appears in 43% of queries. Perplexity processes 200M+ queries/month. Content that AI can't parse is content that doesn't exist.

Try It

npx aeoptimize scan your-site.com

GitHub: https://github.com/dexuwang627-cloud/aeoptimize
npm: https://www.npmjs.com/package/aeoptimize

Star the repo if it helped. Open issues if it didn't.