brain-os-plugin

Analyze skill outcomes + user corrections để propose self-improvements. Triggers: 'improve', 'skill learning', 'why does this skill keep getting it wrong', '/improve [skill-name]'.

Close or merge a pull request the right way via a 4-step hygiene sequence whose order is load-bearing. Use when the user closes, merges, wraps up, lands, or finishes a PR — 'close this PR', 'merge and clean up', 'land this', 'wrap up the PR' — even if they don't spell out all four steps.

Nightly vault maintenance — fix broken wiki-links, detect orphan pages, sync directory indexes, flag stale GH-issue tasks. Use cho: vault maintenance, broken links, orphan pages, nightly cron, lint vault.

Use when mastering a book's content through autonomous extraction, validation against NotebookLM, and knowledge extension

Ultimate autonomous goal→proof loop: auto-grill a fuzzy goal, slice the grounded decisions into a parent PRD + tracer-bullet children, implement each with a TDD evaluator-regenerate loop, then deliver a per-acceptance-criterion PROOF report. Triggers: '/autodev <goal>', 'build this with proof while I'm away', 'autonomously ship and prove this'. For queue-and-walk-away (no drive-to-done, no proof) use /afk; to clarify with a human in the loop use /grill.

Use when verifying ingested book knowledge against NotebookLM, checking verification status, or managing audit flags before absorbing

Turn a goal, task, or existing issue into autonomous work — grill when fuzzy, file/relabel AFK issues, health-check the loop spine, hand off to the schedulers. Add --auto to drive the whole goal→proof loop in one run and return a proof report (delegates to /autodev) instead of queuing-and-handing-off. Triggers: '/afk <goal|task|N>', '/afk --auto <goal>', 'make this AFK', 'finish this while I'm away'. Execute NOW instead → /impl N or cox N.

Aggregate daily working journey across all repos and sessions into structured content material for the '100 days building AI Brain OS' series. Triggers on: journal, daily journey, what did I do today, extract journey, content material

Use when needing to research a topic from the web — scans X and web for state-of-the-art thinking from top voices, synthesizes into bullet points with citations, and optionally ingests into vault

Mode B grill: render ≥4 decision-tree architectures as ASCII (3 axis-aligned + 1 axis-challenger). Use for architecture / system-design topics where recommendations would just get 'ok' — locks in 2-3 turns vs /grill's question-by-question march. Single-decision or pure-exploration → /grill.

Interview the user relentlessly about a topic until reaching a bulletproof, actionable plan. Use when user says 'grill me', 'stress-test this plan', 'poke holes in this', or wants to be challenged on every aspect of a design, strategy, or decision.

Use when entering deep thinking mode, exploring ideas, finding patterns, or reflecting on vault content. Also triggers on: challenge, stress-test, emerge, drift, trace, connect, ideas, ghost, brainstorm, what if, poke holes

Per-issue implementation executor — grab next AFK issue, run /tdd, commit + push + close. Modes: /impl N (specific issue), /impl story <N> (story DAG), /impl auto (drain queue), /impl -p N (parallel workers). Use khi drain AFK backlog hoặc execute approved story.

Workflow-powered autonomous grill — fans out agents to self-answer a topic's decisions from vault/code/web, verifies grounding, and surfaces ONLY the genuine gaps. Use for '/auto-grill <topic>', 'auto-grill this', 'grill this unattended', or as a cron planning front-end. Interactive /grill and /grill-fast are for live human interviews; auto-grill answers itself first.

Use when user wants a briefing on current state — tasks, emails, handovers, priorities. Reads a lazily-regenerated cache. Triggers on: status, what should I do, what's pending, brief me, morning review, wrap up, end of day

Extract knowledge từ Airtable bases vào vault dưới dạng linked entity pages. Sonnet worker + Opus reviewer ≥95% gate, mandatory HITL re-anchor mỗi base mới. Use khi seeding historical knowledge từ company Airtable bases.

Investigate reported bug across code + config + cloud + vault state, file brain-os GH issue với /tdd RED-GREEN fix plan. Use khi skill output sai, cron produces nothing, hook misbehaves, hoặc user nói 'why is X broken'/'debug this'/'/debug'.

Use when testing skill correctness after editing, or wanting to verify all skills pass their eval checks

Execute Gmail triage actions — cleanup inbox (archive, delete, label, star, unsubscribe), draft replies, and process pending triage reports from vault. Use when user says /gmail, wants to clean inbox, process email actions, or draft email replies.

Use when starting a new session and wanting to resume unfinished work from a previous handover

Break grill-session/plan/PRD thành parent story issue + child issues theo tracer-bullet vertical slices. Children có `## Parent` + `## Blocked by`. Use khi convert grill outcome thành AFK tickets, file vào sonthanh/ai-brain. Triggers: 'slice this', '/slice', 'break into issues'.

Use when running the complete knowledge pipeline for a book, from self-learn through ingestion, verification, absorption, and sync

Transcribe YouTube/podcast/audio URL ra clean text. Auto-captions trước, fallback whisper-cpp + Silero VAD. Use trước khi /research video/audio source, hoặc user nói 'transcribe'/'get the transcript'/'get raw quotes from'.

One-shot principle check via advisor. Use when you want to stress-test a decision, plan, or code against your principles — non-interactive, returns findings. For interactive stress-testing, use /grill instead.

Test-driven development for brain-os artifacts via red-green-refactor. Use when building or fixing skills, scripts, hooks, plists, or vault tooling — mention 'red-green-refactor', 'TDD', 'test-first', 'tracer bullet', or invoke with /tdd.

System-level reorganization scan — detects shallow modules, cross-skill duplication, unused code, vault drift. Use khi user nói 'reorganize'/'refactor'/'audit architecture'/'system-level review'. Khác /improve (per-skill) và /vault-lint (mechanical hygiene).

Use when ending a session with unfinished work that needs to be continued in a future session

Use when extracting deep insights from a book and connecting them to vault zones (business, personal, thinking)

Capture key moments mid-session with full context. Triggers on natural language: aha, mark this, key moment, that's interesting, remember this decision, this is important, breakthrough

One-time Gmail setup: scan 3 months of inbox history to build gmail-rules.md with sender patterns, importance signals, and cleanup rules. Use when setting up Gmail automation for the first time or refreshing email rules.

Use when processing raw knowledge input (book notes, highlights, raw files) into structured vault notes