Seo Skills Pack

SEO Skills Pack

Portable SEO operating system for Codex, Claude Code, and Agent Skills-compatible tools.

This is not a pile of prompts.

It is the instruction layer behind my local SEO squad: router memory, specialist skills, practitioner canon, DEJAN reverse-engineering workflows, and Claude-ready subagents that turn retrieved context into useful action.

If seo-vector-snapshot is the memory engine, this repo is the operating system wrapped around it.

Current Refresh

This release adds a clearer live-versus-archive memory model:

• the main live layer is now restricted to the highest-signal SEO and AI-search sources
• the older noisy live-seo-feed-* layer has been removed from the pack
• archive backfill now lives in a separate skills/seo/memory/archive/ folder
• router files were updated so the system knows when to use strict freshness versus broader backfill
• newer Ahrefs, DEJAN, and Hobo-derived notes were added to the library

Why This Repo Feels Different

Most “AI SEO packs” are static.

This one is operational:

• it routes work to the right lens instead of relying on one generic prompt
• it carries practitioner-shaped memory, not just abstract tips
• it installs into real agent runtimes instead of living as copy-paste docs
• it pairs with a portable vector snapshot so the system can remember and act
• it now separates a strict live layer from a slower archive backfill layer so the corpus can grow without polluting the main freshness monitor

What You Get

• the full SEO router and memory library
• the DEJAN AI-search reverse-engineering pack
• Ahrefs and programmatic SEO helper skills
• durable dated memory notes and source-canon notes
• a strict live source layer for high-signal SEO and AI-search monitoring
• an archive backfill layer for secondary reputable sources
• installers for Codex, Claude Code, and generic Agent Skills directories

Platform Support

| Platform | Status | Install Path | | --- | --- | --- | | Codex | Ready | ~/.codex/skills and ~/.codex/memories | | Claude Code / Claude CLI | Ready | ~/.claude/skills, ~/.claude/agents, ~/.claude/CLAUDE.md | | Agent Skills-compatible tools | Ready | any target skills directory | | OpenClaw | Usable through mirrored files and indexing | import or sync workflow |

Snapshot At A Glance

| Metric | Value | | --- | ---: | | Captured | 2026-03-20 | | Skill directories | 5 | | Markdown notes inside skills/seo | 134 | | Markdown files inside skills/dejan-ai-reverse-engineering | 6 | | Archive backfill notes inside skills/seo/memory/archive | 7 | | Durable memory notes | 7 |

Included In The Pack

| Path | Purpose | | --- | --- | | skills/seo | primary SEO router, memory layer, and operating docs | | skills/seo-coral | specialist SEO role pack | | skills/dejan-ai-reverse-engineering | AI-search, grounding, AI Mode, and citation analysis | | skills/ahrefs | backlink and export-driven workflows | | skills/programmatic-seo | scalable page-system and feasibility workflows | | skills/seo/memory/archive | slower archive backfill from secondary reputable sources | | memories/*.md | durable dated notes from the live research system |

Why This Repo Matters

Without a skill layer, a vector DB is just stored text.

This repo adds the missing parts:

• routing
• specialist instructions
• curated memory organization
• portable SEO context on a fresh machine

That means a second laptop does not just have files. It has working SEO judgment scaffolding.

What Makes It Valuable

• turns raw retrieval into operator judgment
• gives Claude, Codex, and other agent runtimes a shared SEO instruction layer
• preserves expert lenses from DEJAN, Glenn Gabe, Marie Haynes, Patrick Stox, Cindy Krum, Lily Ray, Mike King, and others
• makes a fresh machine feel like it already knows your research environment

Built For Ongoing Growth

This pack is no longer just a static export.

It now reflects a tiered memory system:

• a strict live layer for high-signal freshness
• a separate archive layer for slower reputable-source backfill
• router notes that keep the main monitoring layer clean
• a companion vector DB workflow that keeps new notes queryable

Current Content Shape

Strict live layer:

• Google Search Central
• Ahrefs
• DEJAN
• GSQi
• Marie Haynes
• Lily Ray
• MobileMoxie
• Brodie Clark
• iPullRank
• Jono Alderson

Archive backfill layer:

• Hobo
• Aleyda Solis
• Search Engine Land
• Search Engine Journal
• Search Engine Roundtable

Install Into Codex

git clone https://github.com/vijaychauhanseo/seo-skills-pack.git
cd seo-skills-pack
./scripts/install_to_codex.sh

Default install target:

• ~/.codex

Override the target:

CODEX_HOME=/path/to/codex-home ./scripts/install_to_codex.sh

Install Into Claude Code Or Claude CLI

git clone https://github.com/vijaychauhanseo/seo-skills-pack.git
cd seo-skills-pack
./scripts/install_to_claude.sh

This installs:

• user-level skills into ~/.claude/skills
• SEO subagents into ~/.claude/agents
• a user memory import into ~/.claude/CLAUDE.md

After install, Claude can use:

• /seo
• /seo-coral
• /ahrefs
• /programmatic-seo
• /dejan-ai-reverse-engineering

And the installed subagents:

• seo-researcher
• dejan-reverse-engineer

Install Into Any Agent Skills Directory

If another tool supports the Agent Skills open standard, copy the skills into its own skills directory:

./scripts/install_to_agent_skills_dir.sh /path/to/agent/skills

Best Paired Setup

For full portability:

1. Install this repo into ~/.codex
2. Clone seo-vector-snapshot
3. Use the DB with the installed skills for query + routing

For Claude Code:

1. Install this repo with ./scripts/install_to_claude.sh
2. Install seo-vector-snapshot with its Claude installer or MCP installer
3. Use the vector query skills, the MCP server, or let Claude route work to the installed SEO subagents

Example:

SQUAD_MEMORY_SKILLS_ROOT=~/.codex/skills \
python3 ../seo-vector-snapshot/tools/squad_memory.py decide \
  "Need a practitioner for AI Mode, grounding, and GSC reporting" \
  --db ../seo-vector-snapshot/db/squad_memory.db

Best Use Cases

• AI search reverse engineering
• AI Overviews and grounding diagnostics
• practitioner-informed SEO analysis
• programmatic SEO planning
• export-driven Ahrefs workflows
• multi-source SEO memory on a second machine
• growing a large local SEO memory system without turning the main live layer into noise

Why It Is Public

I wanted these repos to show a more useful model for AI-assisted SEO work:

• memory that compounds
• skills that route
• expert context that survives the session
• a stack that moves across tools instead of being trapped in one interface

Not Just Generic SEO Notes

This pack includes:

• practitioner canon from DEJAN, Glenn Gabe, Marie Haynes, Patrick Stox, Cindy Krum, Lily Ray, Brodie Clark, Mike King, and others
• durable notes built from live ingestion and manual curation
• a clean split between high-signal freshness notes and broader archive backfill notes
• role-pack structure so the system knows which lens to apply
• skill files that already match the Agent Skills open standard used by Claude Code

Companion Repository

The retrieval snapshot that pairs with this pack lives here:

• seo-vector-snapshot

Use this repo for skill execution and memory routing. Use the vector snapshot for portable retrieval, MCP tool calling, and query resolution.

Social Preview Asset

If you want a custom GitHub social preview card for this repo, use:

• assets/social-preview.png

Claude Adapter Files

Claude-specific files live under:

• adapters/claude-code/user-memory/
• adapters/claude-code/agents/

License

MIT. See [LICENSE](LICENSE).

Source & license

This open-source MCP server is cataloged on AgentStack and links to its original source — we do not rehost the code.

• Author: vijaychauhanseo
• Source: vijaychauhanseo/seo-skills-pack
• License: MIT

Install and usage instructions live in the source repository linked above.