メニュー
Turns a rough description into a structured GitHub issue with gitmoji title, problem statement, acceptance criteria, optional interview mode, and posting via gh CLI.
Codex または Claude でインストール この Prompt をコピーして Codex、Claude、または他のアシスタントに貼り付けると、Skill ページを確認してインストールできます。
SOC 職業分類に基づく
| name | create-issue |
| description | Turns a rough description into a structured GitHub issue with gitmoji title, problem statement, acceptance criteria, optional interview mode, and posting via gh CLI. |
Turn a rough description into a well-structured GitHub issue. The skill formats your input into a consistent template — gitmoji title, problem statement, acceptance criteria, and more — then posts it with gh issue create. An optional --interview mode reads the codebase first and asks targeted questions before drafting anything.
feat, bug, chore, refactor, perf, docs, test, security, ui) from keywords and prefixes the title with the matching gitmoji--interview mode reads git ls-files, README, and key entry-point files before asking targeted questions--dry-run prints the gh issue create command without executing it--type flag lets you override auto-detectiongh issue create; works with --repo owner/repo for non-current repos/create-issue <description> # direct — build from description
/create-issue # interactive — ask for title + description
/create-issue --interview # interview mode: read repo, ask questions
/create-issue --type feat|bug|chore|refactor|perf|docs|test|security|ui
/create-issue --labels "bug,needs-triage"
/create-issue --repo owner/repo # target a specific repo
/create-issue --dry-run # preview without posting
Read reference/issue-format.md when selecting a gitmoji or rendering the issue body template.
parse args: extract <description>, --type, --labels, --repo, --dry-run, --interview
interview mode (if --interview):
git ls-files to understand project structuredirect / interactive mode:
<description>: ask "Describe the issue:"bug / error / crash / broken / fix → bugadd / implement / support / new / feature → featrefactor / simplify / clean / reorganize → refactorperf / slow / cache / optimize / speed → perfdoc / readme / comment / guide → docstest / spec / coverage → testsecurity / vuln / auth / sanitize / inject → securitystyle / ui / design / layout / css → uichore--type was given, use it; skip auto-detectionbuild template: fill all sections; omit 🔁 Steps to Reproduce for non-bug types; omit 💡 Technical Notes if the user skipped; prefix title with matched gitmoji
show preview: print the full title + body and ask:
Create this issue? [yes / edit / cancel]
post:
gh issue create \
--title "<gitmoji> <title>" \
--body "<body>" \
[--label <labels>] \
[--repo <owner/repo>]
--dry-run: print the command above without running itreport: print the issue URL returned by gh
gh issue create without user approval- [ ] format so GitHub renders them as task itemsgit ls-files + file reads, not git log; keep it fast (2–3 files max)Generates changelog entries from git history following Keep a Changelog conventions and cuts versioned releases. Use when the user wants to update CHANGELOG.md with recent commits or promote [Unreleased] to a versioned release.
Creates a conventional git commit with gitmoji prefix and lowercase imperative message. Use when the user is ready to commit staged changes or wants help composing a clean, atomic commit message.
Scaffolds production-ready CLI projects from language templates, supporting Go with Cobra, Makefile, golangci-lint, goreleaser, and GitHub Pages docs; and Python with Typer, uv, ruff+mypy, PyPI OIDC trusted publishing, and GitHub Pages docs. Use when the user wants to bootstrap a new command-line tool project from scratch.
Analyzes a codebase and generates LLM-optimized documentation covering project overview, architecture, build system, testing, development patterns, deployment, and a file catalog. Use when the user wants to document a codebase or refresh existing project documentation.
Transforms documents into polished self-contained HTML by reading Markdown/text directly, converting other supported inputs with uvx markitdown, selecting from 20 purpose-built templates, and generating the output.
Bootstraps a new agent skill by generating SKILL.md, README.md, LICENSE, and optional reference files as a directory inside the pavelsimo/skills repo and registering it in the index README. Use when the user wants to create a new Claude Code skill from scratch.