seo-vector-snapshot

<p align="center"> </p>

Portable SEO memory graph for AI search, technical SEO, and multi-agent workflows.

This is not a backup of notes.

It is the working retrieval layer behind my local SEO squad: a portable SQLite snapshot, learned routing priors, task-pack logic, and CLI + MCP access for Codex, Claude Code, OpenClaw, and shell-capable agents.

If your SEO workflow touches AI Overviews, grounding, AI Mode, technical audits, or reverse engineering, this repo gives you the part of the system that remembers.

This snapshot now reflects the newer corpus shape behind the squad:

• the DB has been rebuilt against the stricter reputable-source freshness layer
• archive backfill from secondary reputable sources is now represented in the corpus without polluting the main live monitor
• snapshot metadata has been updated to the current graph size and prior counts
• the portable CLI remains intact, so the repo still works off-machine

Most SEO research dies in tabs, loose docs, and half-recalled “I saw this somewhere” moments.

This repo turns that mess into a portable memory layer:

• not a raw vector dump, but a retrieval system with query tooling and task packs
• not generic SEO text, but practitioner-shaped memory built from live research and curation
• not locked to one runtime, but usable through CLI, Claude skills, and native MCP tools
• not another restart-from-zero setup, but a handoff-ready memory graph for the next laptop or operator

• preserves working SEO context instead of forcing fresh research every session
• moves a trained retrieval layer between machines in minutes
• exposes the same memory through multiple access layers: CLI, MCP, Claude skills, and task packs
• pairs with `seo-skills-pack` so retrieval b