jnemargut
- 35 skills
- 0 followers
- 6 hours ago last updated
- ▌ Brief · jnemargutGenerate a shareable one-page HTML project brief from decisions already made. Reads .decisions/ output and produces a polished, self-contained page for stakeholders or teammates. Use after a thinking skill has produced decisions and you want something to share.
- ▌ Observe · jnemargutExtract observations, assumptions and tensions hiding in any artifact — transcripts, strategy docs, proposals, PRFAQs, planning notes, meeting notes, Slack threads. Produces a visual analysis page plus structured data. Use when you have raw material and need to see what's actually in it.
- ▌ Shape · jnemargutGeneral-purpose design and execution skill for any complex goal. Walks through the detailed implementation decisions with visual previews, comparison tables, and recommendations. Works for anything that needs careful design before execution: event planning, content creation, physical projects, organizational redesigns, or any situation with meaningful design choices.
- ▌ Render · jnemargut bundleRender a local HTML or SVG file to a PNG and look at it. Use before showing the user any HTML artifact, decision page, mockup, report, chart, or SVG asset — and whenever a layout, chart, or visual needs checking rather than assuming it worked.
- ▌ Challenge · jnemargutReads your existing decisions and gently challenges them. Surfaces contradictions between decisions, flags weak or missing reasoning, identifies assumptions that haven't been tested, highlights unresolved tensions, and asks the questions you might not be asking yourself. Doesn't change anything - just holds up a mirror.
- ▌ Game Plan · jnemargutGeneral-purpose planning skill that creates a research-backed operational roadmap for any complex goal. Career transitions, legal battles, major life changes, organizational challenges, or anything where you need a concrete phased plan with specific tasks. Each task tagged as requiring you personally, AI assistance, or fully automatable.
- ▌ Journal · jnemargutDecision Journal mode for recording, visualizing, and tracking decisions when you're the expert. You bring the answers, AI helps you see them clearly and remember them. Every decision gets a full visual page. Changes are tracked with reasoning. AI gently challenges weak reasoning once, then moves on.
- ▌ X Shape · jnemargut bundleExperimental v2 of /shape. Same soul — design the detailed execution of any complex goal (events, renovations, content series, weddings, restructures) through visual decisions — plus: an explicit constraints ledger with a running cost/time tally across decisions (it flags when your choices collectively blow the budget), parallel research subagents, calibrated confidence with honest toss-ups, a steelman case against every recommendation, stakes/reversibility triage, and tripwires + pre-mortem in the summary. Reads /whoiam and /research-sources config.
- ▌ Excavate · jnemargutCodebase decision archaeology. Reads an existing codebase and surfaces the invisible decisions encoded in it — from tech stack choices down to approach-level patterns like memory allocation strategies, rendering pipelines, and collision detection. Adapts to any domain (game engines, web apps, CLI tools, embedded systems). Shows evidence, you confirm what's right and correct what's not. Re-run after changes to catch drift.
- ▌ Investigate · jnemargutTurn assumptions into evidence. Takes assumptions from /observe or your own hypotheses, converts the riskiest into testable ones, researches each from multiple angles, and produces insights with recommendations. Pauses twice for human judgment. Use when you have beliefs you need to check before acting on them.
- ▌ X Challenge · jnemargutExperimental v2 of /challenge. Same soul — reads your existing decisions and gently challenges them in conversation (no HTML, changes nothing) — plus: it understands the v2 decision metadata (confidence, case-against, assumptions, tripwires, dissent flags), patrols your tripwires against current reality with fresh web research, re-tests toss-up decisions to see if new information has settled them, revisits decisions where you overrode a strong recommendation, and red-teams its own challenges before showing them so only the ones that survive scrutiny reach you. Supports 'challenge hard' for a sharper pass.
- ▌ X Game Plan · jnemargut bundleExperimental v2 of /game-plan. Same soul — a research-backed phased operational roadmap for any complex goal, every task tagged human/AI-assisted/automatable — plus: parallel research including failure stories, a day-by-day Week One, dependency tags on tasks, per-phase checkpoint gates with pivot criteria, and automatic conversion of your strategy brief's tripwires into recurring check-in tasks. Reads /whoiam and /research-sources config.
- ▌ Product Plan · jnemargut bundleProduct planning skill that generates a research-backed launch roadmap — operations, partnerships, trust-building, product evolution, and go-to-market execution. Each task tagged as human-only, AI-assisted, or automatable. Part of the Product Pipeline: use after /product-strategy and before /product-design.
- ▌ Decide · jnemargutThe entry point to all thinking and action skills. Just say what's on your mind, the orchestrator reads your intent, picks the right skill, and takes you there. Clear inputs get routed instantly. Vague ones get a quick 'did you mean this or that?' then route. You don't need to know which slash command to use. Use when someone says anything that could be handled by a thinking or action skill but doesn't know which one, or when they want a natural-language entry point instead of memorizing slash commands.
- ▌ Strategize · jnemargutGeneral-purpose strategy skill for any complex problem or goal. Researches the domain, identifies the key decisions that matter, and walks you through each one with real research, visual options, and recommendations. Works for career moves, legal situations, personal goals, business challenges, or any situation where you need to think deeply before acting.
- ▌ Whoiam · jnemargutTell the system who you are so it frames decisions in your language, not its own. A lightweight identity skill that captures your role and domain familiarity, then saves a profile that every thinking skill reads to adapt its language, analogies, and comparison dimensions to your expertise level. Two questions max. Takes 15 seconds.
- ▌ X Product Plan · jnemargut bundleExperimental v2 of /product-plan. Same soul — a research-backed launch playbook covering everything outside the code (operations, partnerships, trust, supply seeding, go-to-market), every task tagged human/AI-assisted/automatable, with 'Product at This Phase' capability requirements per phase — plus: parallel research including launch post-mortems, the riskiest-assumption test from your strategy brief as literal Task #1, per-phase go/pivot/kill gates, dependency tags, and tripwire check-in tasks. Part of the Product Pipeline: use after /x-product-strategy, before /x-product-design.
- ▌ X Decide · jnemargutExperimental v2 of /decide — the natural-language entry point to the thinking and action skills, routing to the x- (v2) versions wherever they exist. Say what's on your mind; clear inputs route instantly, vague ones get a quick 'this or that?' then route. Knows the whole x-family (x-strategize, x-shape, x-product-strategy, x-product-plan, x-product-design, x-game-plan, x-visual-design, x-challenge) and falls back to the originals for everything else. Passes /autodecide, /overdecide, /underdecide, and '3 takes' modifiers through.
- ▌ X Strategize · jnemargut bundleExperimental v2 of /strategize. Same soul — research the domain, surface the decisions that matter, walk through each with visual options — plus: parallel research subagents, calibrated confidence on every recommendation (including honest toss-ups), a steelman case AGAINST its own pick, stakes/reversibility triage, assumptions with tripwires, and a pre-mortem in the brief. Reads your /whoiam profile and /research-sources preferences.
- ▌ Design System · jnemargut bundleGenerate a unique AI-driven design system through a guided designer workshop. Walks the user through archetype selection, adjective sliders, reference triage, and signature tension — then generates novel SVG icons, a curated Google Fonts pairing, a semantic palette, and an applied showcase HTML + framework-agnostic bundle that any generator (Claude Code, v0, Lovable) can adopt. Every generated choice must trace to a workshop signal (amplification rules), nothing on the banned-defaults list ships, and the showcase passes a render-and-critique gate. Produces 'design/' folder with showcase.html, tokens.json, tokens.css, icons/, and claude.md. Use when the user wants a unique design system (colors, components, typography, icons) generated through discovery — not another shadcn-looking default. Triggers on 'design system', 'brand identity', 'make it unique', 'curated typography', 'not generic', 'design DNA'.
- ▌ Visual Design · jnemargut bundleRe-skin any existing HTML artifact OR SVG asset with a chosen aesthetic. Post-step aesthetic thinking. For HTML: walks through 6 decisions (tradition, composition, color, type, mood, signature flourish) and rewrites the style block — including CSS-only layout art direction. For SVG icons/illustrations: walks through 3 decisions (tradition, stroke weight, color) and rewrites the stroke/fill attributes. Enforces a banned-defaults list and ends with a render-and-critique gate. Produces a side-by-side styled file + tokens.json sidecar. Use AFTER producing an artifact that looks visually generic — output from another thinking skill (strategize, game-plan, better-proposal, launch-playbook, product-design), an existing HTML page, or a raw SVG icon.
- ▌ Hook Init · jnemargutConnect your decisions to an external service for sync, privacy controls, and project bundling. Walks through provider selection, project config, visibility defaults, and auto-sync preferences. Installs hook scripts from a provider's hook package and sets up credentials. Run once per project.
- ▌ Autodecide · jnemargutLike /decide but auto-picks the recommended option for every decision. Same research, same options, same recommendations — skip the per-decision pause and review everything at the end before the action skill fires. Hidden power-user variant of /decide.
- ▌ Overdecide · jnemargutLike /decide but dials the decision count UP. Routes to the right thinking skill and tells it to surface 8-12 decision points instead of the usual 4-7. For when you want to be really thorough and miss nothing. Hidden power-user variant of /decide.
- ▌ Product Design · jnemargut bundleProduct design skill that walks through technical and UX decisions — framework, database, visual direction, user flows, navigation, component design. Presents options as rich HTML documents with rendered visual previews, comparison tables, and recommendations. Part of the Product Pipeline: use after /product-strategy and /product-plan.
- ▌ Underdecide · jnemargutLike /decide but dials the decision count DOWN. Routes to the right thinking skill and tells it to surface only 2-3 decision points instead of the usual 4-7. For when you just want the most important calls made and the rest defaulted. Hidden power-user variant of /decide.
- ▌ Core Principles · jnemargutDerive tension-based principles (tenets) for a strategy, product, or initiative. Amazon-style 'X over Y' format where both sides are genuinely good — you're declaring which one wins when they conflict. Surfaces the real moral, business, and user tensions hiding in your idea and forces you to take a stance.
- ▌ State Your Case · jnemargutA decision circuit breaker for when AI is building. While making code changes, if AI encounters a judgment call that contradicts prior decisions, needs human input, or is too important to decide silently - it stops, generates a visual decision page explaining WHY, presents options, and waits for the human to decide before continuing.
- ▌ X Visual Design · jnemargut bundleExperimental v2 of /visual-design. Re-skin any existing HTML artifact or SVG with a chosen aesthetic — same 6-step flow (tradition, composition, color, type, mood, flourish) for HTML, 3-step for SVG, same banned-defaults list and critique gate — plus: previews rendered from YOUR artifact's actual content instead of generic samples, a Three Takes mode ('3 takes' spawns parallel agents that each produce a complete styled variant to pick from), an eyes-on critique gate that screenshots and visually inspects the result, and a side-by-side compare page. Use AFTER producing an artifact that looks visually generic.
- ▌ Product Strategy · jnemargut bundleProduct strategy skill that walks through the 'what and why' decisions — problem validation, target users, market positioning, business model, and elevator pitch. Conducts real market research before presenting options. Part of the Product Pipeline: use before /product-plan and /product-design.
- ▌ Self Code Review · jnemargutReview your own code before opening a PR. Reads your changes, the codebase, and any prior implementation decisions, then surfaces the judgment calls you should be making: scope drift, architecture fit, testing adequacy, complexity, and PR readiness. Each dimension gets a visual assessment page.
- ▌ Ticket Breakdown · jnemargutThinking skill for engineers. Takes a Jira ticket (or any task description), reads the codebase for context, and walks through the implementation decisions you'd normally make in your head: scope, approach, testing strategy, PR plan, and risks. Produces an implementation brief you can code from.
- ▌ X Product Design · jnemargut bundleExperimental v2 of /product-design. Same flow — walk through technical and UX decisions with rich HTML pages, visual previews, comparison tables — plus: a Living Preview (one composite mockup of YOUR product that accumulates every choice as you make it), design token export (tokens.css/tokens.json), parallel research subagents, calibrated confidence with honest toss-ups, a steelman case against every recommendation, and a slim core that loads the traditions library and HTML templates on demand instead of all at once.
- ▌ X Product Strategy · jnemargut bundleExperimental v2 of /product-strategy. Same flow — walk the 'what and why' decisions (problem, user, market, business model, pitch) with real market research and visual HTML pages — plus: a multi-angle research sweep run as parallel subagents (competitors, numbers, user complaints, failure post-mortems), calibrated confidence with honest toss-ups, a steelman case against every recommendation, permission to say 'don't build this', a riskiest-assumption test plan, and kill criteria in the brief. Reads /whoiam and /research-sources config. Part of the Product Pipeline: use before /x-product-design.
- ▌ Research Sources · jnemargutConfigure your research trust preferences before running other thinking skills. Walk through 5 decisions: your domain, which source types you trust, specific voices and publications, domain allowlist/blocklist, and recency preferences. Saves to .decisions/sources.json which other thinking skills read to shape their research.