Agentic SEO Skill — source summary
An "LLM-first" SEO analysis skill pack for agent IDEs and AI coding assistants
(by Bhanunamikaze). Delivered via Telegram, ingested 2026-05-29. README in
raw/agentic-seo-skill.md.
What it is
A packaged set of agent capabilities: 16 specialized sub-skills (audit, technical,
content/E-E-A-T, schema, sitemap, images, GEO/AEO for AI search, links, GitHub SEO,
…), 10 specialist agents, and 89 scripts (88 Python + 1 shell) used as
"evidence collectors" (audit_runner.py, pagespeed.py, crawl_audit.py,
validate_schema.py, …). CI checks inventory/reference freshness.
Why it's interesting here
It installs the same skill content in each tool's native format — Claude Code,
Codex, Antigravity, Claude Cowork, Cursor, Windsurf, Continue, Copilot, Cline. That
makes it a clean example of agent-skills: portable markdown capabilities an LLM
agent loads — the same shape that claude-financial-services bundles per agent and
gbrain ships as a 43-skill skillpack. (Minor curiosity: its seo schema sub-skill
does Schema.org/JSON-LD generation — the same vocabulary this wiki types pages with.)
Placement: the SEO domain itself is incidental; the skill-pack form factor is why it's here — an instance of agent-skills (the bridge concept, in research-wiki).
A second, larger instance (2026-08-22)
claude-seo is the same form factor at roughly double the size — 25 sub-skills, 18 agents, 32 commands, 14.8k★ against this pack's 16 / 10 / 89-scripts — and the comparison is useful because the two diverge on everything except the domain. This one is cross-tool, installing the same content in nine harnesses' native formats; claude-seo is Claude-Code-only and spends the saved effort on depth (auto-discovery, 15-way parallel dispatch, 410 tests, a four-tier Google credential ladder). Portability versus depth, chosen explicitly, on identical subject matter — the cleanest pair this spoke has for that trade.