メニュー
Show health of the ~/brain/ vector store — collection counts (brain_raw, brain_wiki), indexed vs unindexed files, watcher daemon status, embedding-model consistency, and privacy-mode state. Use when the user says "/brain-status" or asks "is the brain up to date?"
Codex または Claude でインストール この Prompt をコピーして Codex、Claude、または他のアシスタントに貼り付けると、Skill ページを確認してインストールできます。
SOC 職業分類に基づく
Semantic two-tier retrieval over the personal brain at ~/brain/. Queries brain_wiki (synthesized pages) first, falls back to brain_raw (raw source chunks) if confidence is low, and always reports which layer answered with a confidence score. Use when the user says "/brain-query", "search the brain", "semantic search", or asks a factual question about Harsha, his clients, projects, or any topic covered by the brain where exact slugs aren't known.
Ingest a file from ~/brain/raw/ into the wiki. Extract key concepts and entities, create or update pages in ~/brain/wiki/pages/, add cross-references, append to log.md. Use when the user says "/wiki-ingest", "ingest raw/...", or drops a file in raw/ and asks to process it.
Health check on ~/brain/wiki/. Find orphan pages, broken cross-reference links, stale last_updated dates, missing frontmatter fields, concepts mentioned repeatedly without a dedicated page, and contradictions between pages. Use when the user says "/wiki-lint" or "lint the wiki."
Answer a question strictly from the user's brain at ~/brain/wiki/. Never answer from training memory on topics the wiki covers. Always cite the page used. Use when the user says "/wiki-query", "ask the brain", or asks a factual question about their work, clients, or projects.
Persist a synthesized answer (from /brain-query, /wiki-query, or mid-conversation) as a new wiki page at ~/brain/wiki/pages/. Writes frontmatter (title, last_updated, tags), requires at least one cross-reference, updates index.md, appends to log.md, and embeds the new page into brain_wiki. Use when the user says "/wiki-save", "save that to the brain", "file this as a page," or "remember this for later."
| name | brain-status |
| description | Show health of the ~/brain/ vector store — collection counts (brain_raw, brain_wiki), indexed vs unindexed files, watcher daemon status, embedding-model consistency, and privacy-mode state. Use when the user says "/brain-status" or asks "is the brain up to date?" |
Run via Bash:
cd /Users/HAR5HA/brain && .venv/bin/python -m scripts.status
Show the output verbatim inside a fenced code block.
Escalate the following conditions with a direct recommendation:
cd /Users/HAR5HA/brain && .venv/bin/python -m scripts.reembed_all
pid: NOT RUNNING in output) → suggest:
launchctl unload ~/Library/LaunchAgents/com.harsha.brain-watch.plist
launchctl load ~/Library/LaunchAgents/com.harsha.brain-watch.plist
embedding_models in a collection (warning line in output) → this means dim-mismatch risk; recommend:
cd /Users/HAR5HA/brain && .venv/bin/python -m scripts.reembed_all --wipe
PRIVACY_STRICT=on + OpenAI model (fatal line in output) → tell the user to either unset BRAIN_PRIVACY_STRICT in ~/brain/.env OR switch EMBED_MODEL in scripts/config.py to a local provider.hash drift > 0 → files were edited but not re-embedded (watcher missed or was down). Offer /brain-status follow-up with reembed_all.Always echo the Recovery: footer line from the script output so the user always sees the escape hatch.