meijin
Community@texmeijin · Tokyo, Japan
A progressive web developer, who loves TypeScript, Testing, Software Design
Agent Skills by meijin
Showing 21 vetted skills indexed across 2 GitHub repositories.
TeXmeijinauto-grouped-commit
Analyzes uncommitted git changes and groups them into logical commits with messages.
TeXmeijincentering-judge
Validates UI element alignment in PNG screenshots with pixel-level Python scripts.
TeXmeijinprompt-refiner
Refines vague coding requests into execution-ready prompts for other coding agents.
TeXmeijinclean-ai-writing-style
Reviews and rewrites Japanese AI-generated prose through a structured critique loop.
TeXmeijinyarn-classic-to-pnpm
Migrate Yarn Classic repositories to pnpm and audit transitive dependency version drift.
TeXmeijincloudwatch-logs-insights-query
Write and validate AWS CloudWatch Logs Insights QL queries from raw log samples.
TeXmeijinpj-flow
Organizes long-running projects into persistent memory files and timestamped thread folders across conversation sessions.
TeXmeijincodex-collab-review
Coordinates Claude and Codex CLI in tmux panes to produce consensus-based code reviews.
TeXmeijingoal-template-generator
Generates execution-ready GOAL templates from rough task requests for AI agents.
TeXmeijincommit-push
Commits and pushes only files changed in the current conversation thread.
TeXmeijinggg
Searches the web with current-year queries and cites official sources for verified answers.
TeXmeijinvideo-to-gif
Convert video files and screen recordings into optimized GIFs using ffmpeg.
TeXmeijinimage-redaction
Mask sensitive text in screenshots using pixel-detected line coordinates and ffmpeg drawbox filters.
TeXmeijincodex-exec
Delegates rescue, review, and adversarial-review tasks to the Codex CLI from Claude Code.
TeXmeijinharness-creator
Builds machine-driven verification harnesses with Red/Green status detection for code and operations.
TeXmeijingoal-prompt-dispatch
Executes one queued GOAL prompt file per run from a four-folder filesystem queue.
TeXmeijingoal-prompt-enqueue
Converts requests into self-contained GOAL files queued for unattended autonomous execution.
TeXmeijinghostty-applescript
Generates AppleScript layout automation scripts for Ghostty terminal windows, tabs, and pane splits.
TeXmeijinisis
Investigates issues and vague requests as hypotheses against codebase facts before implementation.
TeXmeijinlayered-sequence-diagram
Generate interactive hierarchical sequence diagrams as single-file HTML.
TeXmeijinlayered-flow-chart
Generates interactive, layered flow charts from codebases as single-file HTML.
Frequently Asked Questions About meijin
FAQPage SchemaWhat tasks can I accomplish with meijin's skills?▼
You can group and commit uncommitted Git changes, run dual-agent code reviews via tmux, migrate Yarn Classic repos to pnpm with lockfile drift audits, write CloudWatch Logs Insights queries, build verification harnesses, convert videos to GIFs, redact sensitive image text, and queue GOAL prompts for deferred autonomous execution.
Who are these skills designed for?▼
They target progressive web developers working in TypeScript who value testing and software design. Typical users are engineers managing Git hygiene, performing dependency migrations, reviewing code with multiple agents, and refining vague requests into execution-ready prompts for coding agents like Claude Code and Codex CLI.
How do the GOAL prompt queue skills work in practice?▼
goal-prompt-enqueue writes high-quality GOAL prompts into ~/.goal-prompt-queue/inbox/ after resolving uncertainty via a Pre-GOAL Fact Pass and Decision Funnel. goal-prompt-dispatch later dequeues one item per run, moving it through inbox, inprogress, and done/error folders, executable from Claude, Codex, or cron triggers.
What prerequisites or dependencies do these skills require?▼
Requirements vary per skill: Git for commit skills, tmux plus Claude Code and Codex CLI for collaborative review, ffmpeg or ImageMagick for video-to-GIF and image redaction, AWS access for CloudWatch Logs Insights, macOS AppleScript for Ghostty layouts, and pnpm/Yarn environments for migration audits.
How does centering-judge verify UI alignment?▼
Instead of relying on subjective visual judgment, it writes a new purpose-built verification script per proposition, defining ROI regions, decision formulas, and debug overlays, then runs pixel-level static analysis on PNG screenshots to determine vertical centering, left-edge alignment, or even spacing.