Devbrother Skills
Five skills for daily development: requirement interviews, AGENTS.md rules, PR review triage and single-file HTML design.
- 94 ★ on the repository
- Last commit
- MIT
Extended access
Why this level
- `autofix` edits code and commits fixes for review comments; without `--auto` it asks before changing anything.
- `agents-md` creates and rewrites rule files in the project.
Install
- Works in
In your terminal, with SkillFoxx CLI
npx skillfoxx add skills/devbrother-skillsDetects the agents on your machine, shows the access level and pins the version.
Other ways to install
Access and checks
How we reviewExtended accessThe tool runs code or makes network calls and only reads external data.
- `autofix` edits code and commits fixes for review comments; without `--auto` it asks before changing anything.
- `agents-md` creates and rewrites rule files in the project.
This is third-party code. Review the repository files before installing.
What it does
A collection of five skills that the author devbrother2024 picked from daily workflows. `deep-interview` asks one question at a time until the goal, scope and completion criteria are clear. `agents-md` analyses the codebase, builds an `AGENTS.md` rule system and links it to `CLAUDE.md`. `autofix` reads unresolved PR review threads, checks every comment against the code and fixes only the justified ones. `advisor-strategy` sends a plan to a stronger model that the user names at call time. `artifact-design` plans palette and typography before a single-file HTML page is built. Each skill installs on its own.
Who it is for. Developers who want ready skills for Claude Code, Codex or Cursor that clarify tasks, set repository rules and triage reviews.
Good fit when
- A request is vague, and the goal and completion criteria must be clarified before coding.
- A project has no `AGENTS.md`, and `CLAUDE.md` should point to one.
- A PR has piled up reviewer comments, and justified ones must be separated from doubtful ones.
- A second opinion from a stronger model is wanted before an approach is fixed.
Not a fit when
- The task is tiny, such as a typo or a setting; an interview adds nothing.
- Instructions in English or Russian are required: skill descriptions and `SKILL.md` texts are in Korean.
- UI is edited inside a React or Vue app; `artifact-design` targets standalone HTML files.
Example request
Run deep-interview: I want to add notifications to the app but do not know where to start.Limitations
The repository description and skill texts are in Korean. `advisor-strategy` does nothing without an explicitly named model and runs only through `/advisor-strategy`. `autofix` works with GitHub PRs: by default, fixes are applied after the user confirms, and `--auto` removes that step. Installation through the `skills` CLI relies on the external skills.sh service. In Cursor, after installing, open Settings, Rules, the Agent Decides section and check that the skills are listed.
How to disable. Delete the skill folder from `~/.claude/skills`, `~/.agents/skills` or `~/.cursor/skills`. A plugin installed through a marketplace is removed with the agent's own tools.
FAQ
How do I install the devbrother2024 skills in Claude Code?
Run `npx skills@latest add devbrother2024/skills` and pick the skills. Another route is `/plugin marketplace add devbrother2024/skills` followed by `/plugin install devbrother-skills@devbrother-skills`.
Can I install just one skill?
Yes. Add `--skill <name>` and `--agent <agent>` to the command, and `--list` shows available skills without installing.
How do I call a skill after installing?
With `/skill-name` in the agent chat. In Claude Code the name appears in autocomplete after typing `/`.
Does the set work in Codex?
Yes. Run `codex plugin marketplace add devbrother2024/skills`, then open `/plugins` and install the set.
Where do the skills live after installing?
For Claude Code in `~/.claude/skills` or the project's `.claude/skills`, for Codex in `~/.agents/skills`, for Cursor in `~/.cursor/skills`.
Discussion
No questions or reviews yet.
Related
SkillsNine JS Mastery skills take a change through phases: from idea and spec to verification, tests, release texts and a docs sync
Twenty skills for reading unfamiliar code, reviewing documents and commits, translating Markdown and OpenSpec-driven development
A skills library for Claude Code, Cursor and Codex: team code review, TDD, stacked PRs and repository readiness scoring
Two skills for Codex and Claude Code: prd keeps a living implementation checklist and review audits designs from the user's side
A skill set against AI slop: UI and typography audits, change review, releases, docs and SEO.
Builder Skills
Skills for agents
A set of small skills from Builder.io for agent work habits: visual plans and diff recaps, docs-first checks, cross-agent review and clean status updates
README in short
Five skills and three install routes are described: the `skills` CLI, Claude Code and Codex plugins, and manual folder copying. Each agent has its own skills directory listed.
README badge
Original SKILL.md
---
name: deep-interview
description: |
모호한 요청을 소크라테스식 질문으로 인터뷰해 실행 가능한 요구사항으로 정리한다.
사용자가 deep-interview, 심층 인터뷰, 요구사항 명확화, 생각 정리를 요청하거나 목표, 범위, 제약, 완료 기준이 흐릿할 때 사용한다.
요청이 이미 구체적이거나 단순 오타, 작은 설정 변경, 테스트 보강처럼 물을 가치가 낮은 작업에는 사용하지 않는다.
argument-hint: "<rough request>"
---
# Deep Interview
모호한 요청을 바로 실행하지 말고, 명확한 요구사항으로 정리한다.
핵심은 질문을 많이 하는 것이 아니라, 가장 큰 불확실성 하나를 골라 한 번에 하나씩 푸는 것이다.
소크라테스식으로 묻는다. 답을 대신 정하기보다 사용자의 암묵적 가정, 선택지, 판단 기준이 드러나게 질문한다.
## 질문 축
아래 순서로 가장 불명확한 축 하나를 고른다:
- 목표
- 범위와 제외 범위
- 제약
- 완료 기준
- 기존 맥락과 영향 범위
코드베이스를 보면 답할 수 있는 질문은 사용자에게 묻지 말고 직접 확인한다.
## 진행 방식
질문은 한 번에 하나만 한다. 가장 큰 불확실성 하나를 고르고, 사용자의 판단 기준이 드러나도록 묻는다.
질문 형식:
```md
현재 이해: {요청과 지금까지 정해진 내용을 필요한 만큼 간결하게 요약}
막힌 결정: {지금 풀어야 하는 가장 중요한 불확실성}
질문: {한 가지 질문}
선택지: {필요할 때만 총 2-3개. 마지막 항목은 `- (직접 입력)`}