← ClaudeAtlas

atlas-knowledgelisted

Create a knowledge base step by step: name, location, a scope written well, the filing mode. Also change a scope, rename, adopt an existing Obsidian vault, or forget one. Use for new knowledge base, create a knowledge base, knowledge base for papers, change the scope, adopt this vault, forget this knowledge base.
nathanaday/claude-atlas · ★ 0 · AI & Automation · score 72
Install: claude install-skill nathanaday/claude-atlas
# Create or change a knowledge base Tools: `atlas`, `vault` on the atlas MCP server. A knowledge base holds sources, entities, and concepts, an inbox for sources to ingest, and the user's `ideas/`. It is the Obsidian vault the user opens. Projects use it; a project uses one, a knowledge base serves many. Call `atlas` first: the knowledge bases that exist, their scopes, and the names already taken. ## Ask, in this order 1. **Name.** It becomes the folder's name. 2. **Location.** The folder the knowledge base goes in, as an absolute or `~` path. There is no default place. Suggest the parent folder of the knowledge bases `atlas` lists, when they share one. Never a folder inside another knowledge base. A folder inside a project is fine when that project alone uses it: its operations then commit into the project's repository. Offer `link` for that project afterwards. 3. **Scope.** Two sentences: what it holds, and what it does not. A project session and its ingest read the scope to know what belongs here, so a scope that overlaps another knowledge base's sends sources to the wrong place. Draft one from the user's words, read the existing scopes back, and adjust until each is distinct. Split a knowledge base only at a trust boundary: work and personal, or one that must stay private and one that may be shared. Few and large beats many and small. 4. **Mode.** `generic` files pages by type; `lyt` keeps atomic notes and Maps of Content. Default `