thejustinwalsh

Use when baking derived assets for three-flatland (alpha hitmasks, normal maps, KTX2 encodes, Slug fonts, sprite atlases), deciding whether to bake at all, wiring the baked → runtime fallback, or authoring a new baker

Use when integrating three-flatland with React Three Fiber — registering classes with extend(), rendering sprites declaratively in JSX, Flatland child routing, post-processing via addEffect, resize/render wiring in useFrame/useThree, or avoiding imperative R3F anti-patterns

Use when authoring styles with @stylexjs/stylex — creating styles via stylex.create, applying with stylex.props, defining tokens with defineVars/defineConsts, building themes with createTheme, or migrating inline-styled / CSS / styled-components / Tailwind code to StyleX. Covers the Do/Don't rules and links to the full authoring + installation references.

Use when authoring, restructuring, splitting, or auditing any docs page or content collection in this repo — choosing what type a page should be, deciding what belongs on it, or checking accuracy, structure, and completeness before a release.

Use when writing or reviewing any user-facing prose for three-flatland — landing-page copy, docs-page intros, blog posts, README intros, release notes, or social posts. Required reading before drafting marketing copy in this repo.

Create Excalidraw diagram JSON files that make visual arguments. Use when the user wants to visualize workflows, architectures, or concepts.

Use when the user wants to make/create/author/write a codemod for a three-flatland breaking change, OR when they want to apply/run/execute/migrate using an existing codemod artifact. Routes between authoring (contributors writing migrations) and applying (consumers running migrations) based on the user's intent.

Use this skill when an AI agent needs to inspect, verify, debug, or profile a live Vite app by running temporary snippets inside the browser page and reading browser logs or captured artifacts. Use for client state after interactions, imported app modules, DOM state, human-like input, canvas/WebGL/Three.js state, screenshots, videos, CPU/network/performance/heap analysis, WebXR/Three.js XR with IWER, and runtime-only behavior without editing app files.

Disable restruct prompt refinement. Prompts pass through to Claude without local LLM preprocessing.

Enable restruct prompt refinement. Re-enables the local LLM preprocessing of prompts.

Generate a minimal, research-backed CLAUDE.md for this project. Analyzes the codebase for non-obvious commands and conventions, interviews you for team knowledge, and produces a concise file that stays under 30 rules.

Manually refine a prompt through the local LLM meta-prompt pipeline. Transforms a casual developer request into a structured, rules-aware prompt using a local Ollama model. Only works when features.refinement is enabled in config.yaml.

Review tool permission decisions across sessions and propose auto-approval rules. Analyzes which tools the user consistently approves, then suggests additions to .restruct/permissions.yaml to eliminate repetitive permission dialogs.

Write an initial .restruct/permissions.yaml with safe defaults so the restruct permit hook can auto-approve common tool use without repeated dialogs.

Install Ollama, pull the right model for your hardware, warm it up, and enable the restruct prompt refinement pipeline. Only runs when features.refinement is enabled in config.yaml.

First-time setup for restruct. Runs rule generation (CLAUDE.md), verify wiring, permission defaults, and — if the feature flag is on — prompt refinement setup.

Use when the user wants to file GitHub issues from a planning document, build spec, brainstorm output, or any structured source. Triggers include "create issues for this spec", "file these as issues", "seed this project on GitHub", "break this plan into issues". Drives issue creation only — sets titles, bodies, acceptance criteria, labels, and parent/sub-issue hierarchy. Does NOT implement, does NOT open PRs, does NOT pick what to work on next.

Use when the user provides a GitHub issue number or URL and asks to implement it end-to-end. Triggers include "implement issue

Use when running as the dispatch recommender for a single GitHub repo. Triggers when invoked by middle-management with a state issue number and rate-limit context. The recommender's sole job is to rewrite the state issue body with a ranked dispatch plan and a needs-human digest, conforming to the agent-queue state issue schema. NEVER writes code, NEVER modifies non-state issues, NEVER opens PRs. If asked for anything other than a recommender run, decline and state this skill is recommender-only.

Use when filing or auditing GitHub issues to check their acceptance criteria demand integration into the running product, not just green unit tests. Defines the integration rubric and drives `mm audit-issues`. Triggers include "audit these issues", "do the acceptance criteria hold up", "harden this issue's requirements", and the second pass inside creating-github-issues. Suggests rewrites; does NOT implement, does NOT silently edit issues.

Create Excalidraw diagram JSON files that make visual arguments. Use when the user wants to visualize workflows, architectures, or concepts.

Use when writing or revising any documentation in this repo — README, docs/ guides, TSDoc on public exports, module-index frontmatter, per-folder CLAUDE.md, or prose the Docs-harvester bot generates. Enforces Diátaxis (every doc is exactly one of tutorial / how-to / reference / explanation), the repo's voice (Google + Microsoft style guides), an LLM-ism blocklist, and the accuracy rule that code samples come from working source. Triggers include "write docs for X", "document this package", "audit the docs", "fix the README", "add a how-to". Authoring + audit only — does NOT change code behavior, does NOT open PRs.