connect0459

@connect0459 source repo

11 skills
0 followers
9 hours ago last updated

▌
Gh Pr Draft · connect0459 bundle
Generates a draft PR body for the current branch and writes it to `.connect0459/gh-pr-draft.md`, then optionally creates a GitHub Draft PR from it. Trigger whenever the user asks to create a PR draft, write a PR description, prepare PR body text, or submit a PR — including phrases like "PR bodyの下書きを作って", "PRの説明文を書いて", "draft PRを書いて", "PRを出して", "create PR draft", "write PR body", "submit a draft PR", "PR下書き", or any variation meaning "prepare or submit the pull request". Also trigger when the user finishes implementation work and asks what to do next in the PR workflow.

0
installs
▌
Swift IOS Review · connect0459
Performs Swift code review for iOS apps. Use whenever reviewing Swift code, providing feedback on a PR diff, or checking Swift files for correctness and stability. Always trigger this skill when asked to review Swift code in an iOS app, especially changes involving Presenters, ViewControllers, ViewModels, Reducers, Stores, NotificationCenter observers, threading/DispatchQueue, or test code (XCTest, Nimble, TCA TestStore, etc.). Also trigger when checking architecture layer boundaries in MVP, MVVM, TCA, or any other pattern.

0
installs
▌
TS Solidjs Review · connect0459
Performs SolidJS code review covering both SolidJS-specific reactivity and general design concerns. Always trigger when asked to review SolidJS code or when `.tsx`/`.jsx` files with SolidJS patterns are presented for review. Trigger on phrases like "review this component", "check my Solid code", "any issues here?", or "SolidJSのコードを見て". Distinguish from React: trigger when `createSignal` / `createEffect` / `createStore` / `<For>` or other SolidJS-specific APIs are visible. Do NOT trigger for React-only code (useState, useEffect without SolidJS APIs).

0
installs
▌
Md Tech Blog Review · connect0459
Reviews personal tech blog posts written for Zenn or similar public platforms. Checks technical accuracy, readability, structure, audience fit, and writing expression while preserving the author's unique voice and style. Trigger when the user asks to review a blog post, wants feedback on a Zenn article, or says phrases like "ブログを見てほしい", "記事のレビューをお願い", "Zennに投稿する前に確認したい", "フィードバックがほしい", "記事チェックして", "review this article", or "check my post". Also trigger when the user shares a markdown file that looks like a tech blog draft — especially one with sections like はじめに / TL;DR / おわりに or similar blog structure.

0
installs
▌
Kotlin Android Review · connect0459
Performs Kotlin code review for Android apps. Use whenever reviewing Kotlin code, providing feedback on a PR diff, or checking Kotlin files for correctness and stability. Always trigger this skill when asked to review Kotlin code in an Android app, especially changes involving ViewModels, Fragments, Activities, LiveData, StateFlow, Coroutines, BroadcastReceiver, or test code (JUnit, Mockk, Espresso, coroutines-test, etc.). Also trigger when checking architecture layer boundaries in MVP, MVVM, MVI, or any other pattern.

0
installs
▌
Md Tech Report Review · connect0459
Reviews technical reports and assessment documents used in professional contexts: project evaluations, work assessments, engineering retrospectives, or technical proposals. Checks structural completeness, evidence quality, persuasiveness, and clarity. Trigger when the user asks to review a technical report, work assessment document, evaluation README, or says phrases like "報告書を見てほしい", "技術資料のレビューをお願い", "提出前に確認したい", "アセスメント資料のフィードバックがほしい", "review this report", "check my assessment document". Also trigger when the user shares a structured markdown with sections resembling 課題/解決手法/成果, Background/Solution/Results, or similar structured report format.

0
installs
▌
Whisper Transcribe Setup · connect0459 bundle
Build the whisper-transcribe Rust binary from source and download the default Whisper model (~809 MB, ggml-large-v3-turbo). Installs the binary to ~/.local/bin/whisper-transcribe and the model to ~/.local/share/whisper-transcribe/models/. Runs English and Japanese smoke tests to verify the install works end-to-end. Trigger whenever the user wants to set up whisper transcription, install the whisper backend, build the Rust binary, download the Whisper model, or prepare for local or YouTube transcription — including phrases like "set up whisper", "install whisper", "build whisper-transcribe", "I need to transcribe audio", "prepare for transcription", or any time the user is missing the binary or model required by youtube-transcribe-local or youtube-transcribe-transcribe-youtube. Idempotent: safe to re-run. Build skips when the binary is current; model download skips when the file already exists; smoke tests always run.

0
installs
▌
Youtube Transcribe Local · connect0459 bundle
Transcribe a local audio or video file using ffmpeg + whisper-transcribe. Accepts any format ffmpeg can decode (mp3, m4a, wav, mp4, mkv, mov, webm, opus, flac, …) — transcodes to 16 kHz mono PCM WAV automatically before passing to the Whisper backend. Default output is an SRT file written next to the input. Use whenever the user wants to transcribe a local media file, generate subtitles, get a transcript from audio/video, or convert speech to text — including phrases like "transcribe this file", "get the transcript", "subtitle this video", "speech to text", "run whisper on", or any time the user references a local audio/video file path and asks for text output. Trigger even when the user doesn't say "whisper" — file extension plus transcription intent is enough. Requires /connect0459-agent-skills:whisper-transcribe-setup to have been run first.

0
installs
▌
Youtube Transcribe Translate · connect0459
Translate a transcript or Markdown article between English and Japanese using Claude. Preserves Markdown structure, code, URLs, and technical terms. Output is a language-tagged file next to the source (e.g. talk.md + --to ja → talk.ja.md). Trigger whenever the user wants to translate a transcript or article — "translate this to Japanese", "translate to English", "EN->JA", "JA->EN", "日本語に訳して", "make a Japanese version of this article", "give me the English version" — and especially right after a transcript or article was produced by youtube-transcribe-transcribe-youtube and the user accepts a translation offer, even without saying the word "translate" (e.g. they reply "yes, Japanese" or "do the JP one"). The --to flag is required; Claude prompts the user to specify it when omitted, rather than guessing — short and mixed-language texts mis-detect frequently.

0
installs
▌
Whisper Transcribe Transcribe · connect0459
Internal transcription backend wrapping the `~/.local/bin/whisper-transcribe` binary. Not intended for direct user invocation — called by `/connect0459-agent-skills:youtube-transcribe-local` to transcribe a 16 kHz mono PCM WAV file into SRT, TXT, VTT, or JSON.

0
installs
▌
Youtube Transcribe Transcribe Youtube · connect0459 bundle
End-to-end YouTube → SRT + Markdown workflow. Downloads audio via yt-dlp, transcribes with whisper-transcribe via ffmpeg, and converts the SRT to a structured Markdown article. Use whenever the user wants to transcribe a YouTube video, get a transcript of a YouTube talk, convert YouTube speech to text, or produce a readable article from a YouTube video — including phrases like "transcribe this YouTube video", "get the transcript of this talk", "YouTubeの文字起こし", "YouTubeの動画を書き起こして", "make an article from this video", or any time the user shares a YouTube URL and asks for text, transcript, or summary. Requires /connect0459-agent-skills:whisper-transcribe-setup to have been run first.

0
installs

11 published skills