菜单
Initialize this claude-second-brain vault. Registers qmd collections and generates vector embeddings. Use when: setting up for the first time, re-registering collections, or re-indexing after a bulk ingest session. Trigger phrases: /setup, setup vault, initialize collections, reindex, register qmd.
用 Codex 或 Claude 帮你安装 复制这段 Prompt，粘贴到 Codex、Claude 或其他助手里，让它检查 Skill 页面并帮你完成安装。
基于 SOC 职业分类
Ingest a new source into the wiki. Reads the source, summarizes it, creates a wiki/sources/ page, updates affected topic and entity pages, flags contradictions, and logs the activity. Trigger phrases: /brain-ingest, ingest [file or URL], add this source, read and file this, process this article/paper/note.
Redesign the qmd schema for this vault — analyze the current wiki, recommend a new set of collections and contexts, update scripts/qmd/setup.ts, then tear down and rebuild the index from scratch. Destructive. Trigger phrases: /brain-rebuild, rebuild the brain, restructure qmd, redesign collections, redo contexts, recompute schema.
Refresh the qmd index for this vault — re-scans all collections for new/changed files and regenerates vector embeddings. Use after a bulk ingest session, after manual file edits, or whenever search feels stale. Trigger phrases: /brain-refresh, refresh the brain, refresh embeddings, re-embed wiki, update qmd index, refresh search index.
Query the wiki to answer a question. Searches semantically, synthesizes an answer with inline citations, and optionally files the answer as a permanent Q&A page. Trigger phrases: /brain-search, what do I know about, search the wiki, find everything about, what does the wiki say about.
Use qmd CLI to manage collections, index documents, search, and query a local semantic knowledge base. Use when the user mentions qmd, searching a collection, indexing files, running queries, hybrid search, BM25, vector search, or managing a document index. Covers: collection add/list/remove/rename, context management, get/multi-get, update, embed, query, search, vsearch, and MCP server commands.
| name | setup |
| description | Initialize this claude-second-brain vault. Registers qmd collections and generates vector embeddings. Use when: setting up for the first time, re-registering collections, or re-indexing after a bulk ingest session. Trigger phrases: /setup, setup vault, initialize collections, reindex, register qmd. |
| argument-hint | Optional: 'reindex' to skip registration and only re-index files |
All commands run from the vault root.
Step 1 — Register collections and contexts
Run:
pnpm qmd:setup
Registers the two core qmd collections (wiki, raw-sources) and their path-level context descriptions. Idempotent — safe to re-run.
Step 2 — Index files and generate embeddings
Run:
pnpm qmd:reindex
Scans all collections for new/changed files and generates vector embeddings. The first run downloads ~2GB of local GGUF models — this will take a while.
Skip step 1 and run only:
pnpm qmd:reindex
Do not re-run after every single file edit — batch it after a session.
Run these three commands to confirm everything is working:
# List registered collections (expect: wiki, raw-sources)
INDEX_PATH=/Users/cfh00805122/.cache/qmd/index.sqlite pnpm dlx @tobilu/qmd collection list
# List registered contexts
INDEX_PATH=/Users/cfh00805122/.cache/qmd/index.sqlite pnpm dlx @tobilu/qmd context list
# Show index status and embedding counts
INDEX_PATH=/Users/cfh00805122/.cache/qmd/index.sqlite pnpm dlx @tobilu/qmd status
Both collections should appear in collection list. status should show non-zero document and embedding counts — if embeddings are 0, re-run step 2.
node and pnpm are managed via mise — ensure mise install and pnpm install have been run before setuppnpm dlx @tobilu/qmd commands fail after setup, re-run step 1 then step 2