First Principles — The ruthless reconstruction engine that follows First Questions. Forces deconstruction of the (now-pruned) problem to its irreducible bedrock truths — physics, mathematics, human nature, verifiable primary evidence — then rebuilds the solution from those atoms upward, rejecting all analogy, "best practices," industry norms, and cargo-cult patterns. Directly implements Elon Musk's "boil things down to the most fundamental truths and then reason up from there" and the 5-step Algorithm for real 10x gains instead of 10% increments. Explicitly positioned as the complement (not duplicate) to first-questions: first-questions is the delete/prune gate; first-principles is the create/rebuild engine. Both are mandatory sequential steps in the truth-seeking harness before any serious architecture or implementation. Produces a living `.grok/first-principles/first-principles.md` artifact that, together with first-questions.md, becomes the session's highest-authority contract. Use after /first-questions h
First Questions — The ruthless pre-work ritual that forces brutal clarity before any code, design, or agent orchestration begins. Directly implements Elon Musk's "Question every requirement" and "The best part is no part" philosophy. Every serious session must start (or restart) here: "What are we actually trying to accomplish?" answered in one unambiguous sentence. Aggressively interrogates every requirement, feature, constraint, and assumption for origin, necessity, inertia, and future regret. Produces a living `.grok/first-questions/first-questions.md` artifact that becomes the session's single source of truth. Use at the absolute beginning of any non-trivial work, when scope feels muddy, before calling /universe-grok, or when someone says "we need to add X". Slash command: /first-questions
One-command *complete* setup of the full enforceable Universe Grok harness in any target project. Produces exact structure: full AGENTS.md, .universe-grok/ (CURRENT-TRUTH..., NAMING-CONVENTION, empty decisions/ + truth/), .grok/rules/naming-convention.md (rule for decisions/truth edits), hooks/ (4 enforcement files + full-mode-active marker). Install itself creates no .grok/. Runtime hooks create .grok/research/ + .grok/universe-grok/ only for research artifacts and activation (documented). After this, SessionStart hook injection + bootstrap protocol are mechanically live.
Universe Grok — the ultimate automatic multi-agent orchestration layer for Grok. Activates on any non-trivial software engineering, research, refactoring, feature development, debugging, architecture, or complex task. When active, the agent MUST use optimal subagent topologies (researchers, tacticians, implementers, verifiers, parallel explorers) with fork_context, resume_from, worktree isolation, and mandatory verification loops. Use when the user wants maximum quality, wants to "push Grok to the limit", says "universe", "universe grok", "full power", "orchestrate", or does serious work without micromanaging agents. Slash command: /universe-grok or /universe