Kotaro Chikuba
Community@mizchi · Tokyo/Japan
Kotaro Chikuba (mizchi) maintains 91 skills spanning deterministic visual regression testing, screenshot-to-HTML markup, Cloudflare/AWS deployment, and multi-language code auditing.
Agent Skills by Kotaro Chikuba
Showing 91 vetted skills indexed across 6 GitHub repositories.
mizchivrt-migration-eval
Evaluates visual equivalence of framework or CSS-library migrations using pixel diffs and computed-style deltas.
mizchimock-markup
Converts design mock images into verified HTML/CSS pages through a deterministic verify-markup loop.
mizchivrt-markup-synth
Generates pixel diffs, token audits, and overflow signals from HTML and screenshots using Playwright.
mizchidynamic-markup
Recreate and verify dynamic web page behavior using deterministic Playwright-based gates for breakpoints, scrolling, animations, and interactions.
mizchid2-slides
Builds slide decks from Markdown with D2 figures compiled into self-contained HTML.
mizchiexplain-with-anim
Generates checked explanatory diagrams and animations from code structure using vlmkit-anim.
mizchiagent-validation-loop
Validates developer tools by running disposable subagents through closed-loop scenarios and fixing reported friction.
mizchid2-diagram
Write and verify D2 architecture diagrams rendered as terminal text with TALA layout.
mizchivrt-regression-watch
Detects visual regressions by comparing vlmkit diff runs against persisted per-viewport summaries.
mizchivrt-visual-diff
Compare two rendered pages and generate a structured Markdown visual diff report.
mizchispec-to-playwright
Generates deterministic Playwright tests from natural-language specs with VRT baselines and self-healing.
mizchimarkup-decompose
Decomposes whole-screen markup tasks into components and routes each phase to verification loops.
mizchiauto-markup
Recreates HTML/CSS markup from target screenshots using deterministic pixel-diff measurement loops.
mizchicomponent-vrt
Runs visual regression tests on individual components via Playwright gallery screenshots.
mizchimarkup-assist
Validates HTML and CSS work with deterministic headless-Chromium gates that report selector-attributed defects.
mizchivrt-css-fix-loop
Iterates a VLM-LLM pipeline to repair deliberately mutated CSS until pixel diff falls below threshold.
mizchiexplanatory-animation
Generates checked animations and diagrams from declarative JSON scenes with vlmkit-anim.
mizchipkfire
Manage build and test workflows in Pkl projects with incremental caching.
mizchivibe-scratch-workflow
Accumulate vibe library definitions in a scratch database and finalize, normalize, and apply them.
mizchidotenvx
Encrypt and load .env files for multi-environment workflows.
mizchijustfile
Analyze and document justfile recipes, dependencies, and GitHub Actions integration.
mizchinix-setup
Set up reproducible Nix dev environments with devbox or pure flake.nix workflows.
mizchiupstream-fix-and-pin
Pin downstream dependencies to upstream git refs and track merge SHAs.
mizchiutels-project-bootstrap
Register utels projects and inject ingest tokens into wrangler secrets.
Frequently Asked Questions About Kotaro Chikuba
FAQPage SchemaWhat tasks can I accomplish with mizchi's skill registry?▼
You can run deterministic visual regression diffs per page or per component, convert screenshots and mock images into verified HTML/CSS, audit frontend security, performance, and dependencies, deploy Cloudflare Workers and AWS ECS blue/green releases, detect duplicate code via AST similarity, and produce checked D2 diagrams and explanatory animations.
Who are these skills designed for?▼
Frontend engineers, platform engineers, and coding-agent operators. Frontend specialists get review scorecards, React re-render analysis, and VRT gates; platform engineers get GitHub Actions tuning, OpenTelemetry instrumentation, and AWS OIDC/ECS deployment patterns; agent builders get prompt evaluation and skill-vetting loops.
Do the visual testing skills require a VLM or paid API key?▼
No. Skills like vrt-markup-synth, auto-markup, and markup-assist are pure DOM plus Playwright plus pixel processing, runnable via npx vlmkit with no API key. The agent's own vision handles reasoning; optional VLM assists exist only in scoped experiments like vrt-css-fix-loop.
How do I make Playwright visual tests deterministic in CI?▼
Use spec-to-playwright, which applies the official init-agents planner/generator to author tests from a natural-language spec, adds a determinism layer keeping screenshots byte-stable, and runs @mizchi/vlmkit-heal to auto-fix drifted tests so results reproduce in CI.
What prerequisites do the markup and diagram skills need?▼
Markup skills need Playwright and a target screenshot or URL pair; component-vrt additionally needs a Playwright component-testing gallery (vanilla, React, or Vue references included). Diagram skills need the D2 compiler with TALA layout or the @mizchi/vlmkit-anim binary for animated scenes.