Skip to content

// the library

You don't run these. MasterMind does.

Just describe what you want. MasterMind recognizes it and applies the right skill automatically. No commands to memorize. Prefer to be explicit? Call any of them by name,“debug this”,“use the code-reviewer”. These are the workflows it draws on; bigger expert roles, like design or code review, run as separate agents.

33 skills·4 agents·a token-saving router · a self-leveling loop · swappable field packs · a private lab

skills/

you just talk, these fire automatically

init

set MasterMind up for your project in a few quick steps

help

the full menu: what it does and how to use each part

build

design → implement → verify → review → learn, end to end

debug

structured, evidence-first debugging: not guessing

performance

measure → find the real bottleneck → fix → verify

qa

prove it works end-to-end; adds tests only if you want them

report

a shareable write-up of a cycle: opt-in, off by default

interview

agree what is being built (and what is not) before code

assessnew

is this idea worth building? go, clarify or stop, with evidence

constitutionnew

the rules every feature is checked against, versioned

living-docsnew

product, business and tech docs that stay current

specifynew

what and why, before how: a spec you can trace to code

blueprintnew

the technical plan, checked against your rules first

breakdownnew

tasks grouped so each user story ships on its own

checklistnew

unit tests for your requirements, one concern at a time

analyzenew

do the spec, plan and tasks agree? before any code

convergenew

is it really done? the code checked against every promise

learn

learn your real stack + check its own assumptions

prototype

a fast throwaway experiment to settle one unknown

route

load only the files a task needs: fewer tokens, faster

prompt

sharpen a vague ask into a clear, AI-ready prompt

signature

learn your team's real coding style → rules it follows

persona

write in a named engineer’s documented public style

explain

AI-friendly docs so any model understands your package

quarantine

a private space for sensitive project data: never published

levelup

capture lessons · refresh standards · bootstrap a field

lint

checks MasterMind's own files for bloat and contradictions

clarify

you did not follow the last answer: re-pitched in plain words or as a picture

deepen

surveys where the design is costing you, ranked, and you pick one

double-check

interrogate a claim before you hand it over

deprecate

remove, migrate, retire: safely and provably

roadmap

a decision map that survives months of work

handoff

a handoff that survives a context reset