🔍 ASO Skill — Agentic Search Optimization

By Capxel — Authors of the LLM-LD open standard | ClawHub

---

AI agents don't search the web — they recommend. ASO ensures your business is the one they recommend.

What It Does

Command	Purpose	Output
audit	Score any website 0–100 on AI readability	Terminal report or JSON
generate	Generate Layer 1 ASO files (JSON-LD + LLM-LD index)	aso-output/<domain>/
compare	Before/after: what AI sees without vs with ASO	Markdown comparison report

Why This Exists

$4.3 trillion in commerce is now influenced by AI-powered recommendations. When someone asks ChatGPT, Perplexity, or Claude to recommend a business — your website needs to be readable by those agents. Most aren't.

ASO (Agentic Search Optimization) was coined by Dominick Luna in November 2025 and is built on the LLM-LD open standard — the structured data format designed specifically for large language models.

This skill gives you the free Layer 1 audit and generation tools. For full 3-layer deployment (knowledge graphs, entity files, agent context blocks, LIGHTHOUSE monitoring), see capxel.com/aso.

Installation

# From ClawHub
clawhub install aso

# From .skill package
openclaw skill install ./aso.skill

# From source
openclaw skill install ./

Usage

Audit a website

# Via OpenClaw (natural language)
"Run an ASO audit on example.com"

# Direct CLI
python3 scripts/audit.py https://example.com
python3 scripts/audit.py https://example.com --json

Generate Layer 1 files

python3 scripts/generate.py https://example.com
python3 scripts/generate.py https://example.com --max-pages 20 --out aso-output

Compare before/after

python3 scripts/compare.py https://example.com

What the Audit Checks

• ✅ Schema.org JSON-LD, Microdata, RDFa
• ✅ LLM-LD index file (/.well-known/llm-index.json)
• ✅ AI subdomain reachability (ai.<domain>)
• ✅ AI Discovery Page (/ai-discovery)
• ✅ LLM Disco Network membership
• ✅ Entity/knowledge graph indicators
• ✅ Meta descriptions, Open Graph, structured contact data

Security

• Zero dependencies — Python stdlib only. No pip install. No supply chain risk.
• No telemetry — No analytics, tracking, or phone-home behavior.
• No API keys required — No credentials, tokens, or environment variables.
• Read-only network — HTTP GET only to user-specified domains.
• Local output — Files written to aso-output/ only.
• Input sanitized — All URLs validated before use.

See full security details in SKILL.md.

Technical Details

• Python: 3.9+ (stdlib only)
• Dependencies: Zero
• Network: HTTP GET via urllib with curl fallback
• User-Agent: OpenClaw-ASO/1.0
• Crawl limit: Default 20 pages, 1s delay between requests

The LLM-LD Standard

LLM-LD (Large Language Model Linked Data) is an open standard for making websites machine-readable to AI agents. It defines:

• Layer 1 — Schema.org JSON-LD (what this skill generates for free)
• Layer 2 — Entity files and knowledge graphs
• Layer 3 — AI discovery index (llm-index.json) with validation

Specification: llmld.org/spec/llm-ld-v1

Links

• LLM-LD Standard: llmld.org
• LLM Disco Network: llmdisco.com
• Full ASO Deployment: capxel.com/aso
• Free ASO Test: capxel.com/aso/test
• Source: github.com/Capxel/aso-skill

License

MIT — see LICENSE.

---

Powered by Capxel — Creators of the LLM-LD Standard