ontology-bootstraplisted

# /ontology-bootstrap — fill an empty vault from the code Two facts make a fresh `oh-my-ontology` vault feel empty: 1. `oh-my-ontology init` only seeds 5 *example* nodes — they're meant to be replaced, not extended. 2. Hand-authoring the first 20–30 nodes is the heaviest friction in the onboarding path (measured: ~25 cli `add` calls in the Paravel real-codebase dogfood — `docs/dogfood-paravel-2026-05-06.md`). This skill closes that gap with **3 MCP calls total** — one read (`analyze_repo_structure`) plus two batch writes (`add_concepts` for the nodes, `add_relations` for the edges). Down from ~25 round-trips. It is the *cold-start* counterpart to `/ontology-sync` (which keeps an already-grown vault in step with new code). ## When to run **Run when** any of these are true: - the user says "이 codebase 분석해줘" / "bootstrap the ontology" / "fill the vault from this repo" / similar. - the user asked you to do anything ontology-related and `list_kinds` shows ≤ 5 nodes (only starters). - the user just ran `oh-my-ontology init` and is asking what to do next. **Skip when**: - the vault already has 20+ user-curated nodes — at that point `/ontology-sync` (incremental) is the right tool. - the user explicitly opted out (e.g. "I'll add nodes by hand") — respect it. - there is no reachable repository (running in a non-code dogfood folder). ## Workflow The MCP server (`oh-my-ontology-mcp`, R16 v0.8.0+) exposes `analyze_repo_structure`. CLI wrapper: `oh-my-ontology analyze [ro