co-evolution-wikilisted
Install: claude install-skill SiyanHub/siyan-brain-skills
# Co-Evolution Second Brain Wiki
A reusable methodology and toolset for building a long-term personal (or small-team) knowledge base where the AI is a guarded co-author, not a free-hand ghostwriter. Distilled from a production second-brain vault that has been running since 2026-07.
The wiki lives as an Obsidian vault. Markdown is the only source of truth. Derived retrieval indexes are disposable and rebuildable. The owner — never the AI — controls when a full update runs and what counts as identity narrative.
## Why this exists
A plain "let the AI dump notes into a folder" wiki drifts within weeks: facts get overwritten, sensitive values leak, identity gets inferred from project output, contradictions get silently resolved. This skill installs the guardrails that keep a second brain trustworthy over years:
- A **three-layer model** separates facts, self-narrative, and structural evolution so the AI cannot rewrite who the owner is.
- A **five-channel entry map** routes every request (read-only query, global deep-dive, task sediment, self-interview, manual full maintenance) to a bounded write surface.
- **Seven judgment types** and a **three-zone write boundary** decide, for every candidate fact, whether it auto-writes, needs confirmation, or must be explicitly approved.
- An **inbox buffer** catches conflicts, sensitive values, ambiguous mappings, and self-narrative material instead of letting them silently land as fact.
- **L0–L3 tiered retrieval** answers questions from