OrbitOS

Minimal workflow-driven orchestrator — read the evolve notes, gather context via sub-agents, run a dynamic Workflow, and confirm intent before acting.

Daily planning workflow - review last note, plan today, connect to active projects

Lightweight sub-agent dispatcher — fast fan-out that skips formal restatement and review-by-default

Review English drafts for grammar, idiom, and naturalness. Correct inline with terse explanations.

Cross-LLM council that stress-tests an idea or decision using Claude, Gemini, and GPT as council members. User relays prompts to Gemini and GPT. Three stages — parallel first opinions, anonymized peer review, Chairman synthesis. Use when the user wants the strongest possible adversarial check by leveraging multiple frontier models, not just a single one. Inspired by karpathy/llm-council.

Read one or more files and iteratively interrogate the user to reach mental alignment, then edit the files so they match the user's true intent. Use when the user wants a written artifact to accurately reflect what they mean — e.g. "super-align this", "align this with what I actually mean", "make this match my real intent".

Review session context and persist actionable lessons into per-skill evolution.md files. Scriptless — Claude uses native Read/Edit/Write tools directly.

Adversarial spaced-repetition review (full deep / quick recall) of Wiki notes; Research/Resources by request

Generate a handoff prompt that transfers the mental model — the WHY and WHAT — so the follow-up AI can own the HOW entirely. TRIGGER when: user says "handoff prompt", "hand this off", "another context-clean AI", "pick up later", "delegate this task", or any variation requesting a transferable prompt for task continuation or delegation.

Create an atomic Wiki note with auto-linking and wikilink discovery

Generate Excalidraw diagrams from natural language descriptions. Use when asked to "create a diagram", "make a flowchart", "visualize a process", "draw a system architecture", "create a mind map", or "generate an Excalidraw file". Supports flowcharts, relationship diagrams, mind maps, and system architecture diagrams. Outputs .excalidraw JSON files that can be opened directly in Excalidraw.

Verify Zombie Roguelike changes by running the TypeScript build and checking the dev server log for errors. Use before claiming a task done.

Use when user asks to analyze, track, summarize, or report on personal expenses from Alipay (支付宝) or WeChat Pay (微信) CSV exports. Produces a single self-contained interactive HTML dashboard with a concise Claude-written analysis section embedded. Runs entirely locally — raw transaction data never enters Claude's context.

Archive completed projects

Meta-skill that switches the main agent into orchestrator mode — decompose, dispatch sub-agents, review, and synthesize

Park one or more loose ideas into today's daily note as flat checkbox captures. Use when the user has thoughts — a new todo, a research angle, a thing to remember — they want filed under today's execution plan rather than triaged to the inbox or the current project note. This is the lightweight flat-capture path, distinct from `/breakdown-tasks`.

Decompose a single already-identified task or goal into a stratified subtask tree with time estimates and blocker markers, so the user can see the full scope before starting and track progress granularly. Use when the user points at one thing they need broken down — distinct from `/daily-note-addition`, which is the flat-capture path for parking multiple loose independent ideas without decomposition.

Converts an idea or an Inbox note into a structured Project Note

On-demand adversarial self-critique.

Evening shutdown workflow - review the day, fill Evening Review, identify deferred tasks, wind down

Complete API for Google NotebookLM - full programmatic access including features not in the web UI. Create notebooks, add sources, generate all artifact types, download in multiple formats. Activates on explicit /notebooklm or intent like "create a podcast about X"

Explain concepts using the Feynman Technique, treating the user as a 12-year-old non-native English speaker. Use this skill whenever the user says "Feynman", "explain like I'm 12", "ELI12", "simple explanation", "teach me like a kid", "feynman-style", "explain it simply", or any phrasing that requests a beginner-friendly, jargon-free breakdown of a concept. Works for any subject — physics, math, CS, machine learning, or everyday topics. Do NOT use for tasks like writing code, generating files, or editing documents.

Insert Feynman-style explanation callouts directly into notes at the point of confusion. Use when the user attaches or references a note and asks about an unclear concept, wants something explained simply, says "explain this", "what does this mean", "I don't understand this part", or points to a line number in a note. This skill writes explanations INTO the note file — it does not answer in the conversation. Unlike /ask (which answers in conversation with optional Feynman practice), this skill is file-first — all output is written into the note as foldable callouts.

Quick answers first, then opt-in Feynman expression practice for conceptual questions

Deep research workflow for technologies, concepts, or complex topics

Guide step-by-step derivation of physical and mathematical formulas

Surface the Game Framework's identity statement, 1-year goal, and daily levers for reflection and discussion

Provide realistic, reasoning-based time estimates for today's tasks

Interactive brainstorming session, then optionally create a Project or capture knowledge