Infinite Claude Code memory: doctrine before Obsidian
Doctrine first (claude-memory-system + structure-avant-outil), Obsidian after. claude-code (base) → claude-memory-system (4 memory levels = the doctrine) → obsidian (human perspective on the same doctrine) → obsidian-skills (automated workflow). Implements the structure-avant-outil pattern.
Why
Infinite memory starts with a doctrine, not a tool. Two weeks looking for that decision: why you picked additive RLS over a separate schema. You dig through git log, commit messages, prompt history: nothing. The decision lived in one session, died with it. That's the infinite-memory myth: it's not a storage problem, it's a what-do-we-store problem. This kit lays out doctrine first, tool second. You accept that not everything deserves to be remembered: trivial bug fixes die in commits, standard features too. Only structuring decisions (the why that won't survive in code) deserve a note. And…
When
Maturity phase, after several months of project, for a dev/PM/founder who has already lost a decision and refuses to repeat it. ✅ You know it's your kit when You hunt for a past decision, and can't find it in git log or commits You make the same call twice: without remembering you'd already settled it You work on multiple projects, and lose the thread between them ❌ Not for this kit You want…
Tools included
- Claude Code
- Claude Memory System
- Obsidian
- Obsidian Skills
- Graphify
- Qmd