B.O.S.S.

Quick guide

Everything, on one page.

You don't memorize any of this. Inside a project, boss map lists what you have and what's one unlock away, and the conscience points at the right verb when it's relevant. This page is for scanning before you install — and for the one week you forget a command.

always works

The commands that don't change

True in every mode. If you remember one, make it boss status.

boss new <name>scaffold a new project (Quickstart mode)
boss adoptbring BOSS into a repo you already started
boss maplive cheatsheet: where you are + what’s one unlock away
boss status [--conscience]mode / pinned version / drift (+ loop states)
boss boardwhat’s in flight (captured → shipped)
boss unlock <mode>climb a rung: quickstart → mvp → v1 → scale
boss team [add @user]who’s on the venture (solo by default)
boss conscience pause --for 8hsilence the whole conscience for a bounded sprint
boss conscience mute <moment>turn down ONE moment; unmute to bring it back
boss removetake BOSS back out — preview first, --apply to do it
/boss-syncpull the latest BOSS practices into this project
/boss-learnroute a proven pattern UP into the library or DOWN into the app
reach for

I want to…

…get an idea out of my head /boss <idea> to spin up · /triage to capture without committing · /import <file|url> if it's already written down
…find out whether it's real /canvas pressure-tests it · /persona works out who it's for · /comp-eval finds out who else is already solving it · /interview preps the conversation with a human · /pretotype builds a fake door · /evidence records what you actually learned
…see something move today /prototype — the smallest clickable version to react to
…build the thing properly /spec writes the feature down · /smoke is the gate · /log and /close end the loop
…keep the design from drifting /design-tokens-init lays the tokens · /design-library generates the gallery and the reuse index from your code (V1) · /design-review and /ux-check are the review pass
…not get hacked or embarrass myself /red-team (add --humane to probe your own product for dark patterns) · /ai-failure-states · /trust
…put it in front of someone /landing writes the page · /ship deploys it · /onboard is the first-run experience
…know if it's working /measure · /retention · /money · /health
…remember what I decided /decide writes it down with a falsifier · boss board shows what's in flight
…ask someone who's done this /consult convenes the mentors · or address one directly, e.g. mentor-customers
…stop /sunset — ending a project well is a skill, and BOSS treats it as one
by mode

Every skill, and the rung it arrives on

Generated from the stage manifests, current as of v0.198.0. If this page and the tool ever disagree, the tool is right and the release gate is broken.

Quickstart 16 skills

/welcome First time using BOSS? Start here /welcome
/boss Spin up a freshly-scaffolded project from a rough idea or PRD /boss [path-to-PRD | rough idea text]
/import Bring existing material into this project — point at a file, a folder, or a URL (Word doc, Google Doc, Obsidian note, PDF, slide deck, online reference) and BOSS pulls a durable copy into docs/source/ and folds it into your idea /import <path-or-url> [more paths/urls] [IDEA-NNN]
/triage Capture an idea — and keep adding to it /triage <thought> (run again anytime to add more)
/prototype Get the idea out of your head and onto the screen — fast /prototype [IDEA-NNN | rough idea text]
/canvas Pressure-test an idea as a humane business, and show the same answers to whoever needs to read them /canvas [IDEA-NNN] [--frame humane|lean|bmc]
/evidence Capture one piece of evidence about your riskiest assumption as a durable EVID-NNN record — paste your notes or describe what happened, and BOSS drafts the file with an HONEST grade on a fixed 3-rung ladder (stated-pain → observed-behavior → commitment) /evidence [paste notes | describe what happened]
/interview The bridge from the conscience's best advice ("a 15-minute call with the right person beats another canvas pass") to captured evidence /interview [prep | debrief | paste your notes]
/research Turn a real research transcript into product context and graded evidence /research [transcript file | paste the transcript]
/decide Record a load-bearing decision as a durable DEC-NNN record — Context / Decision / Why / a cheap Falsifier (what would prove this wrong, and by when) / Consequences, stamped with who decided (founder vs AI-suggested-ratified vs AI-autonomous) and how reversible it is /decide <the decision, or describe it>
/sunset End something honestly — a whole project, or one zombie feature validating it's actually dead (not just quiet), guarding the segment/commitment exception (a low-usage feature can be the reason a key account stays), and drafting the HONEST user-facing deprecation notice (real notice period, a path out, no "we're improving your experience" euphemism). Subtraction is a feature; a bloated product serves no one. Framed as an experiment that returned an answer, never as failure. Deliberate-invoke only; the conscience never suggests it unprompted (it may only point at it from inside a moment that already fired — e.g. the focus circuit-breaker). Usage - /sunset [FEAT-NNN | feature name]
/persona Build your app's target-user persona from your idea, grow it from what you know + online research + any real user research you drop in, and consult it as an agent voice — both to guide product decisions ("would my user want this?") and to QA your builds ("how would she react to this screen?") /persona [derive | enrich <slug> | consult <slug> "question"]
/comprehend Read what BOSS can actually understand about this project (the captured idea, the source material, or the adopted repo) and tell the founder where they stand — POSITION, never a grade - where the work is, what BOSS can't see, two or three options with what each would change about their week, and what changes in how they work /comprehend
/feedback Send feedback about BOSS itself back to the people who build it — a bug, a confusion, a wish, a "this got in my way." User-initiated and transparent: it shows you exactly what it will send (and the small bit of context attached) before anything leaves your machine, then files it as a GitHub issue upstream (or gives you a prefilled link to paste) /feedback [what's on your mind]
/boss-sync Pull current BOSS practices into this project — bring the installed modes' skills/agents up to the latest version as a reviewed, narrated diff, then bump the project's BOSS pin /boss-sync
/boss-learn Route a proven pattern two ways — UP into the BOSS library as a reusable superset practice, or DOWN into this app as hardened core functionality /boss-learn [what to promote]

MVP 29 skills

/spec Promote an idea into a buildable spec — IDEA-NNN becomes FEAT-NNN with a goal, acceptance criteria, a smoke check, and the three paths that must not break (the money path, the destructive path, and the negative path — who must NOT be able to see this) /spec [IDEA-NNN] (or describe the feature inline)
/smoke Run the project's smoke check — "is the app even working right now?" Stack-configured /smoke
/log Append a dated entry to docs/devlog.md — what landed this session, what's next, what surprised you /log <one-line summary or detailed entry>
/close Session-end ritual — update docs/RESUME.md (state + next tasks + open decisions), append a /log entry, and let the conscience update its read on the venture (.boss/brain/) /close
/evals Build and run the eval set for an AI-mediated FEAT — "is it correct?" paired with /smoke's "is it alive?" Husain's discipline applied to LLM-mediated control-flow in your app — look at your data, build the eval set FIRST, categorize failures by mode, vibes-based eval is only a starting point /evals [FEAT-NNN | --new <feat>]
/pretotype Test demand BEFORE you build /pretotype [IDEA-NNN]
/design-tokens-init Scaffold the minimal three-layer design token system at the first UI commit /design-tokens-init
/ai-cost Establish AI spend discipline for your app — declare per-user / per-feature / monthly budgets, name the model choices, wire a per-call cost logger, set a review cadence /ai-cost
/cost-review Read the AI cost ledger and produce a dated review /cost-review
/ai-first-init Bake the AI-first discipline into your app from day one /ai-first-init
/ai-failure-states Design what your app does when the AI fails — the five failure states (garbage / refusal / hallucination / timeout / cost-spike) and the declared response for each /ai-failure-states
/extract Pause and sort patterns — PRINCIPLE #1's discipline as a skill /extract
/drift-deep The deep, whole-project version of the conscience's drift check — "am I fooling myself across EVERYTHING I've built?" Reads the entire project (canvas, all devlog, all FEAT specs, the actual code, the ideas) and judges, honestly, whether the body of work is validating the named riskiest assumption or building around it /drift-deep
/revalidate The 3-line gate before paused work re-enters the build — checks a deferred idea/feature against a world that moved (still relevant? still aligned? anything changed?) and routes it to revive / rescope / kill / re-pause, so you never build a zombie feature /revalidate [ID or paused item]
/judge-traces Error analysis on your real session traces — the Hamel/Shankar discipline applied to your own work /judge-traces [last N | all]
/consult Convene the mentor board on a cross-cutting question — route it to the mentors who actually have a stake, get each one's take in their own lens, and synthesize the answer with the disagreements kept visible (not averaged away) /consult <question>
/red-team Adversarially test an AI-mediated FEAT (or BOSS's own conscience hook, --self) against the OWASP LLM Top 10 — and, when the target is an agent (tools + memory + autonomy), the OWASP Agentic ASI Top 10 (Dec 2025) — tool misuse, agentic supply chain, memory poisoning, and the rest /red-team [FEAT-NNN | --paths | --self | --humane]
/practice Capture a craft learning — a better way to build with AI you found — as a shared, attributed PRAC-NNN record your cofounder gets too /practice <what you learned>
/ship Put your app where a real user can hit it — the CD half of building /ship [--preview | --rollback]
/landing Generate the founder's FIRST landing page — on-brand, honest, out the block to scale /landing [--demand | --product]
/measure The post-ship counterpart to /pretotype /measure
/onboard Design the path from signup to the aha-moment — the highest-leverage number in the funnel (activation > acquisition; fix it and every downstream cohort lifts at once) /onboard
/roadmap Weigh what to build next — feedback against behavior — into a SMALL bet-list with a mandatory NO-list /roadmap
/trust Earn trust honestly — the AI-specific privacy/compliance load-bearing set, without the SOC2 theater /trust
/health The post-launch verdict AND the diagnosis, in one read - is this working, and if it isn't, where exactly is it breaking? Renders product-market fit honestly (pre / at / post, defaulting to pre-PMF because scaling before fit is the #1 way startups die), and when the curve is decaying, says WHERE it dies — at the top (activation), in the middle (engagement), or at the wallet (failed payments) — and routes to the one real fix /health
/money The money verb, from the first dollar to operating it /money [upgrade | dunning | price-raise | expansion]
/design-review Before-code design review for your app /design-review [FEAT-NNN | path-to-component-spec]
/ux-check After-code UX review for your app /ux-check [route-or-component-path | FEAT-NNN]
/comp-eval Find out who else is solving this, and keep knowing /comp-eval [<space or problem> | add <name> | recheck [<name>]]

V1 2 skills

/board The cross-FEAT sequencing surface for your app /board (or /board --next, /board --blocked, /board --by-cohort)
/design-library Generate the visual design library for your app — a self-contained HTML gallery of foundations (color/type/spacing), components (every variant, all five states) AND the rule sets (principles, do/don't pairs, terminology, voice), rendered from the code and tokens so it cannot drift /design-library [--check]

Scale 1 skills

/incident The blameless one-page post-mortem for an outage — /sunset's honest-accounting shape, scoped to something that broke in production /incident [what broke]