codex-arsenal 0.1.0 → 0.2.1
This diff represents the content of publicly available package versions that have been released to one of the supported registries. The information contained in this diff is provided for informational purposes only and reflects changes between package versions as they appear in their respective public registries.
- package/CODEX.md +120 -120
- package/README.md +118 -115
- package/lib/installer.js +16 -10
- package/lib/manifest.js +10 -1
- package/package.json +1 -1
- package/skills/publishing-npm-packages/SKILL.md +158 -0
package/CODEX.md
CHANGED
|
@@ -1,121 +1,121 @@
|
|
|
1
|
-
# CODEX.md
|
|
2
|
-
|
|
3
|
-
> Behavioral guidelines to reduce common LLM coding mistakes.
|
|
4
|
-
> Merge with project-specific instructions as needed.
|
|
5
|
-
> **Tradeoff:** These guidelines bias toward caution over speed. For trivial tasks, use judgment.
|
|
6
|
-
|
|
7
|
-
---
|
|
8
|
-
|
|
9
|
-
## 1. Think Before Coding
|
|
10
|
-
|
|
11
|
-
**Don't assume. Don't hide confusion. Surface tradeoffs.**
|
|
12
|
-
|
|
13
|
-
Before implementing:
|
|
14
|
-
|
|
15
|
-
- State your assumptions explicitly. If uncertain, ask.
|
|
16
|
-
- If multiple interpretations exist, present them — don't pick silently.
|
|
17
|
-
- If a simpler approach exists, say so. Push back when warranted.
|
|
18
|
-
- If something is unclear, **stop**. Name what's confusing. Ask.
|
|
19
|
-
|
|
20
|
-
---
|
|
21
|
-
|
|
22
|
-
## 2. Simplicity First
|
|
23
|
-
|
|
24
|
-
**Minimum code that solves the problem. Nothing speculative.**
|
|
25
|
-
|
|
26
|
-
- No features beyond what was asked.
|
|
27
|
-
- No abstractions for single-use code.
|
|
28
|
-
- No "flexibility" or "configurability" that wasn't requested.
|
|
29
|
-
- No error handling for impossible scenarios.
|
|
30
|
-
- If you write 200 lines and it could be 50, rewrite it.
|
|
31
|
-
|
|
32
|
-
> Ask yourself: *"Would a senior engineer say this is overcomplicated?"*
|
|
33
|
-
> If yes — simplify.
|
|
34
|
-
|
|
35
|
-
---
|
|
36
|
-
|
|
37
|
-
## 3. Surgical Changes
|
|
38
|
-
|
|
39
|
-
**Touch only what you must. Clean up only your own mess.**
|
|
40
|
-
|
|
41
|
-
When editing existing code:
|
|
42
|
-
|
|
43
|
-
- Don't "improve" adjacent code, comments, or formatting.
|
|
44
|
-
- Don't refactor things that aren't broken.
|
|
45
|
-
- Match existing style, even if you'd do it differently.
|
|
46
|
-
- If you notice unrelated dead code, **mention it — don't delete it**.
|
|
47
|
-
|
|
48
|
-
When your changes create orphans:
|
|
49
|
-
|
|
50
|
-
- Remove imports / variables / functions that **YOUR changes** made unused.
|
|
51
|
-
- Don't remove pre-existing dead code unless asked.
|
|
52
|
-
|
|
53
|
-
> **The test:** Every changed line should trace directly to the user's request.
|
|
54
|
-
|
|
55
|
-
---
|
|
56
|
-
|
|
57
|
-
## 4. Goal-Driven Execution
|
|
58
|
-
|
|
59
|
-
**Define success criteria. Loop until verified.**
|
|
60
|
-
|
|
61
|
-
Transform tasks into verifiable goals:
|
|
62
|
-
|
|
63
|
-
| Vague | Verifiable |
|
|
64
|
-
|-------|------------|
|
|
65
|
-
| "Add validation" | "Write tests for invalid inputs, then make them pass" |
|
|
66
|
-
| "Fix the bug" | "Write a test that reproduces it, then make it pass" |
|
|
67
|
-
| "Refactor X" | "Ensure tests pass before and after" |
|
|
68
|
-
|
|
69
|
-
For multi-step tasks, state a brief plan:
|
|
70
|
-
|
|
71
|
-
```
|
|
72
|
-
1. [Step] → verify: [check]
|
|
73
|
-
2. [Step] → verify: [check]
|
|
74
|
-
3. [Step] → verify: [check]
|
|
75
|
-
```
|
|
76
|
-
|
|
77
|
-
Strong success criteria let you loop independently.
|
|
78
|
-
Weak criteria ("make it work") require constant clarification.
|
|
79
|
-
|
|
80
|
-
---
|
|
81
|
-
|
|
82
|
-
## 5. Agentic Execution (Codex-specific)
|
|
83
|
-
|
|
84
|
-
**Codex runs with more autonomy — so guardrails matter more.**
|
|
85
|
-
|
|
86
|
-
- **Sandbox first:** Never run destructive commands (`rm -rf`, DB migrations, deploys) without explicit confirmation.
|
|
87
|
-
- **Read before write:** Understand the existing codebase structure before modifying it. Use `find`, `grep`, `cat` to build context.
|
|
88
|
-
- **One action at a time:** In agentic loops, complete and verify each step before proceeding to the next.
|
|
89
|
-
- **Fail loudly:** If a step fails, surface the error immediately. Don't try to work around it silently.
|
|
90
|
-
- **Scope your shell:** Prefer relative paths over absolute. Never assume a working directory without confirming it.
|
|
91
|
-
|
|
92
|
-
---
|
|
93
|
-
|
|
94
|
-
## 6. Context Awareness
|
|
95
|
-
|
|
96
|
-
**Codex has a limited context window — use it wisely.**
|
|
97
|
-
|
|
98
|
-
- Don't re-read files you've already loaded. Reference prior output instead.
|
|
99
|
-
- When summarizing large codebases, compress systematically: directory tree → key files → function signatures.
|
|
100
|
-
- If context is filling up, state it: *"I'm approaching context limits. Summarizing and continuing."*
|
|
101
|
-
- Prefer targeted grep/search over reading entire files when looking for a specific symbol.
|
|
102
|
-
|
|
103
|
-
---
|
|
104
|
-
|
|
105
|
-
## 7. Tool Use Discipline
|
|
106
|
-
|
|
107
|
-
**Every tool call has a cost. Make them count.**
|
|
108
|
-
|
|
109
|
-
- Batch related reads into a single operation where possible.
|
|
110
|
-
- Don't run the same command twice to "verify" — if you need to verify, define the assertion first.
|
|
111
|
-
- Avoid speculative exploration ("let me just check...") — know what you're looking for before you look.
|
|
112
|
-
- If a tool fails, diagnose before retrying. Don't brute-force.
|
|
113
|
-
|
|
114
|
-
---
|
|
115
|
-
|
|
116
|
-
## ✅ These guidelines are working if:
|
|
117
|
-
|
|
118
|
-
- Diffs contain fewer unnecessary changes
|
|
119
|
-
- Fewer rewrites due to overcomplication
|
|
120
|
-
- Clarifying questions come **before** implementation, not after mistakes
|
|
1
|
+
# CODEX.md
|
|
2
|
+
|
|
3
|
+
> Behavioral guidelines to reduce common LLM coding mistakes.
|
|
4
|
+
> Merge with project-specific instructions as needed.
|
|
5
|
+
> **Tradeoff:** These guidelines bias toward caution over speed. For trivial tasks, use judgment.
|
|
6
|
+
|
|
7
|
+
---
|
|
8
|
+
|
|
9
|
+
## 1. Think Before Coding
|
|
10
|
+
|
|
11
|
+
**Don't assume. Don't hide confusion. Surface tradeoffs.**
|
|
12
|
+
|
|
13
|
+
Before implementing:
|
|
14
|
+
|
|
15
|
+
- State your assumptions explicitly. If uncertain, ask.
|
|
16
|
+
- If multiple interpretations exist, present them — don't pick silently.
|
|
17
|
+
- If a simpler approach exists, say so. Push back when warranted.
|
|
18
|
+
- If something is unclear, **stop**. Name what's confusing. Ask.
|
|
19
|
+
|
|
20
|
+
---
|
|
21
|
+
|
|
22
|
+
## 2. Simplicity First
|
|
23
|
+
|
|
24
|
+
**Minimum code that solves the problem. Nothing speculative.**
|
|
25
|
+
|
|
26
|
+
- No features beyond what was asked.
|
|
27
|
+
- No abstractions for single-use code.
|
|
28
|
+
- No "flexibility" or "configurability" that wasn't requested.
|
|
29
|
+
- No error handling for impossible scenarios.
|
|
30
|
+
- If you write 200 lines and it could be 50, rewrite it.
|
|
31
|
+
|
|
32
|
+
> Ask yourself: *"Would a senior engineer say this is overcomplicated?"*
|
|
33
|
+
> If yes — simplify.
|
|
34
|
+
|
|
35
|
+
---
|
|
36
|
+
|
|
37
|
+
## 3. Surgical Changes
|
|
38
|
+
|
|
39
|
+
**Touch only what you must. Clean up only your own mess.**
|
|
40
|
+
|
|
41
|
+
When editing existing code:
|
|
42
|
+
|
|
43
|
+
- Don't "improve" adjacent code, comments, or formatting.
|
|
44
|
+
- Don't refactor things that aren't broken.
|
|
45
|
+
- Match existing style, even if you'd do it differently.
|
|
46
|
+
- If you notice unrelated dead code, **mention it — don't delete it**.
|
|
47
|
+
|
|
48
|
+
When your changes create orphans:
|
|
49
|
+
|
|
50
|
+
- Remove imports / variables / functions that **YOUR changes** made unused.
|
|
51
|
+
- Don't remove pre-existing dead code unless asked.
|
|
52
|
+
|
|
53
|
+
> **The test:** Every changed line should trace directly to the user's request.
|
|
54
|
+
|
|
55
|
+
---
|
|
56
|
+
|
|
57
|
+
## 4. Goal-Driven Execution
|
|
58
|
+
|
|
59
|
+
**Define success criteria. Loop until verified.**
|
|
60
|
+
|
|
61
|
+
Transform tasks into verifiable goals:
|
|
62
|
+
|
|
63
|
+
| Vague | Verifiable |
|
|
64
|
+
|-------|------------|
|
|
65
|
+
| "Add validation" | "Write tests for invalid inputs, then make them pass" |
|
|
66
|
+
| "Fix the bug" | "Write a test that reproduces it, then make it pass" |
|
|
67
|
+
| "Refactor X" | "Ensure tests pass before and after" |
|
|
68
|
+
|
|
69
|
+
For multi-step tasks, state a brief plan:
|
|
70
|
+
|
|
71
|
+
```
|
|
72
|
+
1. [Step] → verify: [check]
|
|
73
|
+
2. [Step] → verify: [check]
|
|
74
|
+
3. [Step] → verify: [check]
|
|
75
|
+
```
|
|
76
|
+
|
|
77
|
+
Strong success criteria let you loop independently.
|
|
78
|
+
Weak criteria ("make it work") require constant clarification.
|
|
79
|
+
|
|
80
|
+
---
|
|
81
|
+
|
|
82
|
+
## 5. Agentic Execution (Codex-specific)
|
|
83
|
+
|
|
84
|
+
**Codex runs with more autonomy — so guardrails matter more.**
|
|
85
|
+
|
|
86
|
+
- **Sandbox first:** Never run destructive commands (`rm -rf`, DB migrations, deploys) without explicit confirmation.
|
|
87
|
+
- **Read before write:** Understand the existing codebase structure before modifying it. Use `find`, `grep`, `cat` to build context.
|
|
88
|
+
- **One action at a time:** In agentic loops, complete and verify each step before proceeding to the next.
|
|
89
|
+
- **Fail loudly:** If a step fails, surface the error immediately. Don't try to work around it silently.
|
|
90
|
+
- **Scope your shell:** Prefer relative paths over absolute. Never assume a working directory without confirming it.
|
|
91
|
+
|
|
92
|
+
---
|
|
93
|
+
|
|
94
|
+
## 6. Context Awareness
|
|
95
|
+
|
|
96
|
+
**Codex has a limited context window — use it wisely.**
|
|
97
|
+
|
|
98
|
+
- Don't re-read files you've already loaded. Reference prior output instead.
|
|
99
|
+
- When summarizing large codebases, compress systematically: directory tree → key files → function signatures.
|
|
100
|
+
- If context is filling up, state it: *"I'm approaching context limits. Summarizing and continuing."*
|
|
101
|
+
- Prefer targeted grep/search over reading entire files when looking for a specific symbol.
|
|
102
|
+
|
|
103
|
+
---
|
|
104
|
+
|
|
105
|
+
## 7. Tool Use Discipline
|
|
106
|
+
|
|
107
|
+
**Every tool call has a cost. Make them count.**
|
|
108
|
+
|
|
109
|
+
- Batch related reads into a single operation where possible.
|
|
110
|
+
- Don't run the same command twice to "verify" — if you need to verify, define the assertion first.
|
|
111
|
+
- Avoid speculative exploration ("let me just check...") — know what you're looking for before you look.
|
|
112
|
+
- If a tool fails, diagnose before retrying. Don't brute-force.
|
|
113
|
+
|
|
114
|
+
---
|
|
115
|
+
|
|
116
|
+
## ✅ These guidelines are working if:
|
|
117
|
+
|
|
118
|
+
- Diffs contain fewer unnecessary changes
|
|
119
|
+
- Fewer rewrites due to overcomplication
|
|
120
|
+
- Clarifying questions come **before** implementation, not after mistakes
|
|
121
121
|
- Agentic runs complete tasks end-to-end without silent failures or scope creep
|
package/README.md
CHANGED
|
@@ -1,173 +1,176 @@
|
|
|
1
1
|
# Codex-Arsenal
|
|
2
2
|
|
|
3
|
-
|
|
4
|
-
> 플러그인 · 스킬 · 워크플로 · 설정 · 프롬프트 큐레이션 레포지토리.
|
|
3
|
+
Practical building blocks for working with OpenAI Codex, Claude Code, and other coding agents.
|
|
5
4
|
|
|
6
|
-
|
|
5
|
+
Codex-Arsenal is not an awesome list. It is a small, installable collection of agent guidelines, skills, prompts, workflows, configs, and plugin sketches that can be copied into real projects.
|
|
7
6
|
|
|
8
|
-
##
|
|
7
|
+
## Install
|
|
9
8
|
|
|
10
|
-
|
|
11
|
-
"vanilla setup만으로는 부족하다"고 느낀 개발자들을 위한 실용 자료 모음입니다.
|
|
9
|
+
Run the CLI without installing it globally:
|
|
12
10
|
|
|
13
|
-
|
|
14
|
-
|
|
15
|
-
|
|
16
|
-
- setup → 사용법 → 주의사항까지 한 곳에서 해결
|
|
17
|
-
- 개인 개발자 / 소규모 팀 모두를 대상으로 함
|
|
18
|
-
- Claude Code 스타일의 agentic workflow에 최적화
|
|
11
|
+
```bash
|
|
12
|
+
npx codex-arsenal list
|
|
13
|
+
```
|
|
19
14
|
|
|
20
|
-
|
|
21
|
-
> AI 코딩 도구가 처음이라면 [공식 문서](https://platform.openai.com/docs/guides/code)부터 시작하세요.
|
|
15
|
+
Install the default item set into the current project:
|
|
22
16
|
|
|
23
|
-
|
|
17
|
+
```bash
|
|
18
|
+
npx codex-arsenal init --yes
|
|
19
|
+
```
|
|
24
20
|
|
|
25
|
-
|
|
21
|
+
Install specific items:
|
|
26
22
|
|
|
27
|
-
```
|
|
28
|
-
codex-arsenal
|
|
29
|
-
├── plugins/ # 에이전트 기능을 확장하는 플러그인 모음
|
|
30
|
-
├── skills/ # 재사용 가능한 에이전트 스킬 패키지
|
|
31
|
-
├── prompts/ # 프롬프트 템플릿 및 system prompt 예제
|
|
32
|
-
├── workflows/ # 워크플로 정의 및 실행 스크립트
|
|
33
|
-
├── configs/ # .codex, VSCode, zsh, tmux 등 설정 파일
|
|
34
|
-
├── references/ # 논문 · 블로그 · 주목할 GitHub 레포 링크
|
|
35
|
-
├── examples/ # 실전 데모 및 before/after 예제
|
|
36
|
-
└── README.md
|
|
23
|
+
```bash
|
|
24
|
+
npx codex-arsenal get codex-md skill-publishing-npm-packages
|
|
37
25
|
```
|
|
38
26
|
|
|
39
|
-
|
|
27
|
+
Install into another directory:
|
|
40
28
|
|
|
41
|
-
|
|
29
|
+
```bash
|
|
30
|
+
npx codex-arsenal get codex-md --dir ./my-project
|
|
31
|
+
```
|
|
42
32
|
|
|
43
|
-
|
|
33
|
+
On Windows or when running from a package directory with the same name, this form is the most reliable:
|
|
44
34
|
|
|
45
|
-
|
|
35
|
+
```bash
|
|
36
|
+
npm exec --yes --package=codex-arsenal -- codex-arsenal list
|
|
37
|
+
```
|
|
46
38
|
|
|
47
|
-
|
|
39
|
+
## CLI
|
|
48
40
|
|
|
49
|
-
|
|
50
|
-
|
|
51
|
-
|
|
52
|
-
|
|
53
|
-
|
|
54
|
-
| repo-memory | 프로젝트 구조를 기억하는 메모리 레이어 | 계획 중 |
|
|
41
|
+
```bash
|
|
42
|
+
codex-arsenal init [--yes] [--dir <path>]
|
|
43
|
+
codex-arsenal list
|
|
44
|
+
codex-arsenal get <id...> [--dir <path>]
|
|
45
|
+
```
|
|
55
46
|
|
|
56
|
-
|
|
47
|
+
- `init` opens a small selector. With `--yes`, it installs default items without prompting.
|
|
48
|
+
- `list` prints all installable manifest entries grouped by category.
|
|
49
|
+
- `get` installs one or more manifest entries by id.
|
|
57
50
|
|
|
58
|
-
|
|
51
|
+
The package is published on npm as [`codex-arsenal`](https://www.npmjs.com/package/codex-arsenal).
|
|
59
52
|
|
|
60
|
-
|
|
53
|
+
## What Is Included
|
|
61
54
|
|
|
62
|
-
|
|
63
|
-
|------|------|
|
|
64
|
-
| `debug-workflow` | 에러 로그 → 원인 분석 → 수정 → 재현 테스트까지 파이프라인화 |
|
|
65
|
-
| `pr-reviewer` | PR diff를 받아 리뷰 코멘트를 자동 생성 |
|
|
66
|
-
| `test-gen` | 함수 시그니처와 코드를 보고 유닛 테스트 자동 생성 |
|
|
67
|
-
| `refactor-pipeline` | 코드 품질 분석 → 리팩터링 제안 → 적용까지 단계 처리 |
|
|
68
|
-
| `doc-gen` | 코드베이스에서 README, API 문서 자동 작성 |
|
|
69
|
-
| `arch-analysis` | 레포 전체 구조를 분석해 의존성 맵 생성 |
|
|
55
|
+
### Behavior Guidelines
|
|
70
56
|
|
|
71
|
-
|
|
57
|
+
| ID | Installs | Purpose |
|
|
58
|
+
| --- | --- | --- |
|
|
59
|
+
| `codex-md` | `CODEX.md` | Project-local behavioral guardrails for Codex-style agents. |
|
|
60
|
+
| `claude-md` | `CLAUDE.md` | Project-local behavioral guardrails for Claude Code. |
|
|
72
61
|
|
|
73
|
-
###
|
|
62
|
+
### Configs
|
|
74
63
|
|
|
75
|
-
|
|
64
|
+
| ID | Installs | Purpose |
|
|
65
|
+
| --- | --- | --- |
|
|
66
|
+
| `config-codex` | `.codex/config.json` | Starter Codex config. |
|
|
67
|
+
| `config-vscode` | `.vscode/settings.json` | VS Code settings for agent-assisted development. |
|
|
76
68
|
|
|
77
|
-
|
|
78
|
-
- `settings.json` — VSCode + Codex 통합 설정
|
|
79
|
-
- `system-prompts/` — 역할별 system prompt 템플릿 (solo dev / 팀 / 레거시 분석)
|
|
80
|
-
- `.zshrc` 스니펫 — Codex CLI alias 및 단축키 모음
|
|
81
|
-
- `tmux.conf` — AI 세션을 병렬로 띄우기 위한 tmux 레이아웃
|
|
69
|
+
### Prompts
|
|
82
70
|
|
|
83
|
-
|
|
71
|
+
| ID | Installs | Purpose |
|
|
72
|
+
| --- | --- | --- |
|
|
73
|
+
| `prompt-solo-dev` | `prompts/system-prompts/solo-dev.md` | Solo developer system prompt template. |
|
|
84
74
|
|
|
85
|
-
###
|
|
75
|
+
### Skills
|
|
86
76
|
|
|
87
|
-
|
|
77
|
+
| ID | Installs | Purpose |
|
|
78
|
+
| --- | --- | --- |
|
|
79
|
+
| `skill-debug-workflow` | `skills/debug-workflow/` | Reproduce, diagnose, test, and fix bugs systematically. |
|
|
80
|
+
| `skill-test-gen` | `skills/test-gen/` | Generate focused tests from behavior notes and function signatures. |
|
|
81
|
+
| `skill-publishing-npm-packages` | `skills/publishing-npm-packages/SKILL.md` | Prepare, troubleshoot, and automate npm releases with Trusted Publishing. |
|
|
88
82
|
|
|
89
|
-
|
|
90
|
-
|----------|------|---------------|
|
|
91
|
-
| `solo-dev-loop` | 혼자 개발하는 개발자 | 계획 → 코딩 → 테스트 → 커밋을 에이전트와 함께 |
|
|
92
|
-
| `startup-mvp` | 빠른 프로토타입이 필요한 팀 | 스펙 → 구현 → 배포 최단 경로 |
|
|
93
|
-
| `legacy-onboarding` | 오래된 코드베이스 합류 | 에이전트로 구조 파악 → 코드 맵 생성 |
|
|
94
|
-
| `ai-pair-programming` | 주니어/미드 개발자 | 에이전트를 시니어처럼 활용하는 패턴 |
|
|
95
|
-
| `multi-agent-orch` | 복잡한 대형 태스크 | 여러 에이전트에게 역할 분배 후 결과 통합 |
|
|
83
|
+
### Plugins
|
|
96
84
|
|
|
97
|
-
|
|
85
|
+
| ID | Installs | Purpose |
|
|
86
|
+
| --- | --- | --- |
|
|
87
|
+
| `plugin-context-window-compressor` | `plugins/context-window-compressor/` | Plugin sketch for compressing long agent context into concise handoff notes. |
|
|
98
88
|
|
|
99
|
-
###
|
|
89
|
+
### Workflows
|
|
100
90
|
|
|
101
|
-
|
|
91
|
+
| ID | Installs | Purpose |
|
|
92
|
+
| --- | --- | --- |
|
|
93
|
+
| `workflow-solo-dev-loop` | `workflows/solo-dev-loop/` | A plan, code, test, commit loop for solo developers using agents. |
|
|
102
94
|
|
|
103
|
-
|
|
104
|
-
- 주목할 GitHub 레포 (with star 기준 및 실용성 코멘트)
|
|
105
|
-
- 블로그 · 튜토리얼 (검증된 것만)
|
|
106
|
-
- 프롬프트 엔지니어링 컬렉션
|
|
95
|
+
## Repository Layout
|
|
107
96
|
|
|
108
|
-
|
|
97
|
+
```text
|
|
98
|
+
codex-arsenal/
|
|
99
|
+
bin/ CLI entrypoint
|
|
100
|
+
lib/ manifest, installer, and fetcher
|
|
101
|
+
configs/ reusable editor and agent configs
|
|
102
|
+
plugins/ plugin sketches
|
|
103
|
+
prompts/ system prompt templates
|
|
104
|
+
skills/ reusable agent skills
|
|
105
|
+
workflows/ repeatable agentic workflows
|
|
106
|
+
references/ curated references and notes
|
|
107
|
+
examples/ before/after examples
|
|
108
|
+
test/ Node test suite
|
|
109
|
+
```
|
|
109
110
|
|
|
110
|
-
|
|
111
|
+
`lib/manifest.js` is the source of truth for installable items. Add new content there when you want it to appear in `codex-arsenal list` or be installable through `codex-arsenal get`.
|
|
111
112
|
|
|
112
|
-
|
|
113
|
+
## Development
|
|
113
114
|
|
|
114
|
-
|
|
115
|
-
|------|------|
|
|
116
|
-
| ✅ 실사용 검증 | 실제로 사용해봤거나, 커뮤니티에서 반복 검증된 것 |
|
|
117
|
-
| ✅ 실무 적용 가능 | 장난감 예제가 아닌 실제 코드베이스에 적용 가능 |
|
|
118
|
-
| ✅ 반복 사용성 | 한 번만 쓰고 마는 것이 아닌, 루틴으로 쓸 수 있는 것 |
|
|
119
|
-
| ✅ setup 대비 효율 | 설치/설정 비용보다 얻는 생산성이 명확히 큰 것 |
|
|
120
|
-
| ✅ 한계 보완 | Codex/Claude Code의 알려진 약점을 보완하는 것 |
|
|
115
|
+
Run tests:
|
|
121
116
|
|
|
122
|
-
|
|
117
|
+
```bash
|
|
118
|
+
npm test
|
|
119
|
+
```
|
|
123
120
|
|
|
124
|
-
|
|
121
|
+
Preview the npm package contents:
|
|
125
122
|
|
|
126
123
|
```bash
|
|
127
|
-
|
|
128
|
-
|
|
129
|
-
cd codex-arsenal
|
|
124
|
+
npm pack --dry-run
|
|
125
|
+
```
|
|
130
126
|
|
|
131
|
-
|
|
132
|
-
cp configs/vscode/settings.json ~/.vscode/settings.json
|
|
127
|
+
Try the local CLI:
|
|
133
128
|
|
|
134
|
-
|
|
135
|
-
|
|
129
|
+
```bash
|
|
130
|
+
node bin/cli.js list
|
|
131
|
+
node bin/cli.js get codex-md --dir ./tmp-install
|
|
136
132
|
```
|
|
137
133
|
|
|
138
|
-
|
|
134
|
+
## Publishing
|
|
139
135
|
|
|
140
|
-
|
|
136
|
+
This repository is configured for npm Trusted Publishing through GitHub Actions.
|
|
141
137
|
|
|
142
|
-
|
|
138
|
+
Release flow:
|
|
143
139
|
|
|
144
|
-
|
|
140
|
+
```bash
|
|
141
|
+
npm test
|
|
142
|
+
npm pack --dry-run
|
|
143
|
+
npm version patch
|
|
144
|
+
git push --follow-tags
|
|
145
|
+
```
|
|
145
146
|
|
|
146
|
-
|
|
147
|
-
- before / after 생산성 차이가 명확한 것
|
|
148
|
-
- setup 과정이 step-by-step으로 정리된 것
|
|
149
|
-
- 특정 상황(레거시, 대형 레포, 팀 협업 등)에서 효과적인 것
|
|
147
|
+
Use `minor` instead of `patch` when adding new installable content or CLI behavior:
|
|
150
148
|
|
|
151
|
-
|
|
149
|
+
```bash
|
|
150
|
+
npm version minor
|
|
151
|
+
git push --follow-tags
|
|
152
|
+
```
|
|
152
153
|
|
|
153
|
-
|
|
154
|
-
2. 해당 카테고리 폴더에 파일 추가 (각 폴더의 `TEMPLATE.md` 참고)
|
|
155
|
-
3. PR 제목에 카테고리 명시: `[plugin] context-window-compressor 추가`
|
|
156
|
-
4. PR 본문에 **사용 배경 / 실제 효과 / 주의사항** 포함
|
|
154
|
+
The publish workflow runs on `v*` tags and publishes with OIDC, so it does not require a long-lived `NPM_TOKEN`.
|
|
157
155
|
|
|
158
|
-
|
|
156
|
+
## Contribution Guidelines
|
|
159
157
|
|
|
160
|
-
|
|
158
|
+
Good additions should be:
|
|
161
159
|
|
|
162
|
-
|
|
163
|
-
|
|
160
|
+
- useful in real projects, not just demos;
|
|
161
|
+
- small enough to understand quickly;
|
|
162
|
+
- documented with setup and usage notes;
|
|
163
|
+
- installable through the manifest when appropriate;
|
|
164
|
+
- verified with tests or a concrete manual check.
|
|
164
165
|
|
|
165
|
-
|
|
166
|
-
실제로 쓸 수 있는 도구와 패턴을 기록하는 공간입니다.
|
|
166
|
+
When adding a new installable item:
|
|
167
167
|
|
|
168
|
-
|
|
168
|
+
1. Add the files under the appropriate top-level directory.
|
|
169
|
+
2. Add a manifest entry in `lib/manifest.js`.
|
|
170
|
+
3. Run `npm test`.
|
|
171
|
+
4. Run `node bin/cli.js list` and confirm the item appears.
|
|
172
|
+
5. Run `npm pack --dry-run` and confirm the intended files are included.
|
|
169
173
|
|
|
170
|
-
##
|
|
174
|
+
## License
|
|
171
175
|
|
|
172
|
-
|
|
173
|
-
`llm-workflow` `productivity` `automation` `vscode` `terminal`
|
|
176
|
+
MIT
|
package/lib/installer.js
CHANGED
|
@@ -16,28 +16,31 @@ async function readSourceFile(src) {
|
|
|
16
16
|
}
|
|
17
17
|
}
|
|
18
18
|
|
|
19
|
-
export async function installItems(items, targetDir) {
|
|
19
|
+
export async function installItems(items, targetDir, opts = {}) {
|
|
20
20
|
const resolvedTarget = path.resolve(targetDir);
|
|
21
21
|
let installed = 0;
|
|
22
22
|
let failed = 0;
|
|
23
|
+
const successes = [];
|
|
23
24
|
const failures = [];
|
|
25
|
+
const readSource = opts.readSourceFile || readSourceFile;
|
|
24
26
|
|
|
25
27
|
for (const item of items) {
|
|
26
28
|
for (const file of item.files) {
|
|
27
29
|
const destination = path.join(resolvedTarget, file.dest);
|
|
28
30
|
try {
|
|
29
|
-
const content = await
|
|
31
|
+
const content = await readSource(file.src);
|
|
30
32
|
await mkdir(path.dirname(destination), { recursive: true });
|
|
31
33
|
await writeFile(destination, content, "utf8");
|
|
32
34
|
installed += 1;
|
|
35
|
+
successes.push(file.dest);
|
|
33
36
|
} catch (error) {
|
|
34
37
|
failed += 1;
|
|
35
|
-
failures.push({ item: item.id, file: file.src, error: error.message });
|
|
38
|
+
failures.push({ item: item.id, file: file.src, dest: file.dest, error: error.message });
|
|
36
39
|
}
|
|
37
40
|
}
|
|
38
41
|
}
|
|
39
42
|
|
|
40
|
-
return { installed, failed, failures, targetDir: resolvedTarget };
|
|
43
|
+
return { installed, failed, successes, failures, targetDir: resolvedTarget };
|
|
41
44
|
}
|
|
42
45
|
|
|
43
46
|
function defaultItems() {
|
|
@@ -75,14 +78,17 @@ export async function runInit(opts = {}) {
|
|
|
75
78
|
|
|
76
79
|
if (!selected.length) {
|
|
77
80
|
console.log("No items selected.");
|
|
78
|
-
return { installed: 0, failed: 0, failures: [], targetDir: path.resolve(opts.dir || ".") };
|
|
81
|
+
return { installed: 0, failed: 0, successes: [], failures: [], targetDir: path.resolve(opts.dir || ".") };
|
|
79
82
|
}
|
|
80
83
|
|
|
81
|
-
const result = await installItems(selected, opts.dir || process.cwd()
|
|
82
|
-
|
|
83
|
-
|
|
84
|
-
|
|
85
|
-
}
|
|
84
|
+
const result = await installItems(selected, opts.dir || process.cwd(), {
|
|
85
|
+
readSourceFile: opts.readSourceFile
|
|
86
|
+
});
|
|
87
|
+
for (const file of result.successes) {
|
|
88
|
+
console.log(`installed ${file}`);
|
|
89
|
+
}
|
|
90
|
+
for (const failure of result.failures) {
|
|
91
|
+
console.log(`failed ${failure.dest} (${failure.error})`);
|
|
86
92
|
}
|
|
87
93
|
console.log(`\nDone: ${result.installed} installed${result.failed ? `, ${result.failed} failed` : ""}.`);
|
|
88
94
|
return result;
|
package/lib/manifest.js
CHANGED
|
@@ -1,4 +1,4 @@
|
|
|
1
|
-
export const REPO = "
|
|
1
|
+
export const REPO = "citron03/Codex-Arsenal";
|
|
2
2
|
export const BRANCH = "main";
|
|
3
3
|
export const BASE_URL = `https://raw.githubusercontent.com/${REPO}/${BRANCH}`;
|
|
4
4
|
|
|
@@ -60,6 +60,15 @@ export const MANIFEST = [
|
|
|
60
60
|
{ src: "skills/test-gen/prompt.md", dest: "skills/test-gen/prompt.md" }
|
|
61
61
|
]
|
|
62
62
|
},
|
|
63
|
+
{
|
|
64
|
+
id: "skill-publishing-npm-packages",
|
|
65
|
+
category: "Skills",
|
|
66
|
+
label: "publishing-npm-packages",
|
|
67
|
+
description: "A Codex skill for preparing, troubleshooting, and automating npm package releases.",
|
|
68
|
+
files: [
|
|
69
|
+
{ src: "skills/publishing-npm-packages/SKILL.md", dest: "skills/publishing-npm-packages/SKILL.md" }
|
|
70
|
+
]
|
|
71
|
+
},
|
|
63
72
|
{
|
|
64
73
|
id: "plugin-context-window-compressor",
|
|
65
74
|
category: "Plugins",
|
package/package.json
CHANGED
|
@@ -0,0 +1,158 @@
|
|
|
1
|
+
---
|
|
2
|
+
name: publishing-npm-packages
|
|
3
|
+
description: Use when preparing, troubleshooting, or automating npm package releases, especially GitHub Actions Trusted Publishing, provenance, 2FA publish errors, version tags, npm pack checks, or npx CLI packages.
|
|
4
|
+
---
|
|
5
|
+
|
|
6
|
+
# Publishing npm Packages
|
|
7
|
+
|
|
8
|
+
## Overview
|
|
9
|
+
|
|
10
|
+
Publish npm packages through a verified release path. Prefer GitHub Actions Trusted Publishing with OIDC over long-lived `NPM_TOKEN` secrets.
|
|
11
|
+
|
|
12
|
+
## Release Decision
|
|
13
|
+
|
|
14
|
+
Use this order:
|
|
15
|
+
|
|
16
|
+
1. Existing package with Trusted Publishing configured: release by version tag.
|
|
17
|
+
2. First publish for a new package: create the package with npm 2FA or a short-lived/granular manual path, then configure Trusted Publishing.
|
|
18
|
+
3. CI provider unsupported or private dependency constraints: use a granular npm token with the narrowest publish scope.
|
|
19
|
+
|
|
20
|
+
## Preflight
|
|
21
|
+
|
|
22
|
+
Before changing release config, inspect:
|
|
23
|
+
|
|
24
|
+
- `package.json`: `name`, `version`, `bin`, `files`, `repository.url`, `publishConfig.access`
|
|
25
|
+
- `package-lock.json`: present when workflow uses `npm ci`
|
|
26
|
+
- `.github/workflows/*.yml`: CI and publish workflow
|
|
27
|
+
- npm package state: `npm info <package-name>` when npm is available
|
|
28
|
+
|
|
29
|
+
For GitHub Actions Trusted Publishing, require:
|
|
30
|
+
|
|
31
|
+
```yaml
|
|
32
|
+
permissions:
|
|
33
|
+
contents: read
|
|
34
|
+
id-token: write
|
|
35
|
+
```
|
|
36
|
+
|
|
37
|
+
Use GitHub-hosted runners, Node 22.14+ or 24, npm 11.5.1+, and `actions/setup-node` with `registry-url: "https://registry.npmjs.org"`.
|
|
38
|
+
|
|
39
|
+
## Package Metadata
|
|
40
|
+
|
|
41
|
+
Keep npm metadata boring and exact:
|
|
42
|
+
|
|
43
|
+
```json
|
|
44
|
+
{
|
|
45
|
+
"type": "module",
|
|
46
|
+
"bin": {
|
|
47
|
+
"package-name": "bin/cli.js"
|
|
48
|
+
},
|
|
49
|
+
"repository": {
|
|
50
|
+
"type": "git",
|
|
51
|
+
"url": "git+https://github.com/OWNER/REPO.git"
|
|
52
|
+
},
|
|
53
|
+
"publishConfig": {
|
|
54
|
+
"access": "public"
|
|
55
|
+
}
|
|
56
|
+
}
|
|
57
|
+
```
|
|
58
|
+
|
|
59
|
+
Do not use `./` in `bin` paths if npm reports that it auto-corrected the script name or path. Run `npm pkg fix` when available, or apply the equivalent minimal metadata fix.
|
|
60
|
+
|
|
61
|
+
## Trusted Publishing Setup
|
|
62
|
+
|
|
63
|
+
Configure on npmjs.com after the package exists:
|
|
64
|
+
|
|
65
|
+
```text
|
|
66
|
+
Packages -> PACKAGE -> Settings -> Trusted publishing
|
|
67
|
+
Provider: GitHub Actions
|
|
68
|
+
Owner: GitHub owner
|
|
69
|
+
Repository: GitHub repository
|
|
70
|
+
Workflow filename: publish.yml
|
|
71
|
+
Environment: npm, if the workflow uses environment: npm
|
|
72
|
+
```
|
|
73
|
+
|
|
74
|
+
The workflow filename is only `publish.yml`, not `.github/workflows/publish.yml`.
|
|
75
|
+
|
|
76
|
+
If the workflow uses `environment: npm`, ensure GitHub has an environment named `npm`.
|
|
77
|
+
|
|
78
|
+
## Recommended Workflows
|
|
79
|
+
|
|
80
|
+
Use separate CI and publish workflows.
|
|
81
|
+
|
|
82
|
+
CI:
|
|
83
|
+
|
|
84
|
+
```yaml
|
|
85
|
+
name: CI
|
|
86
|
+
on:
|
|
87
|
+
pull_request:
|
|
88
|
+
push:
|
|
89
|
+
branches: [main]
|
|
90
|
+
permissions:
|
|
91
|
+
contents: read
|
|
92
|
+
jobs:
|
|
93
|
+
test:
|
|
94
|
+
runs-on: ubuntu-latest
|
|
95
|
+
steps:
|
|
96
|
+
- uses: actions/checkout@v6
|
|
97
|
+
- uses: actions/setup-node@v6
|
|
98
|
+
with:
|
|
99
|
+
node-version: "24"
|
|
100
|
+
package-manager-cache: false
|
|
101
|
+
- run: npm ci
|
|
102
|
+
- run: npm test
|
|
103
|
+
- run: npm pack --dry-run
|
|
104
|
+
```
|
|
105
|
+
|
|
106
|
+
Publish:
|
|
107
|
+
|
|
108
|
+
```yaml
|
|
109
|
+
name: Publish to npm
|
|
110
|
+
on:
|
|
111
|
+
push:
|
|
112
|
+
tags: ["v*"]
|
|
113
|
+
permissions:
|
|
114
|
+
contents: read
|
|
115
|
+
id-token: write
|
|
116
|
+
jobs:
|
|
117
|
+
publish:
|
|
118
|
+
runs-on: ubuntu-latest
|
|
119
|
+
environment: npm
|
|
120
|
+
steps:
|
|
121
|
+
- uses: actions/checkout@v6
|
|
122
|
+
- uses: actions/setup-node@v6
|
|
123
|
+
with:
|
|
124
|
+
node-version: "24"
|
|
125
|
+
registry-url: "https://registry.npmjs.org"
|
|
126
|
+
package-manager-cache: false
|
|
127
|
+
- run: npm ci
|
|
128
|
+
- run: npm test
|
|
129
|
+
- run: npm pack --dry-run
|
|
130
|
+
- run: npm publish
|
|
131
|
+
```
|
|
132
|
+
|
|
133
|
+
## Release Procedure
|
|
134
|
+
|
|
135
|
+
1. Verify clean state: `git status --short`.
|
|
136
|
+
2. Run tests: `npm test`.
|
|
137
|
+
3. Inspect package contents: `npm pack --dry-run`.
|
|
138
|
+
4. Bump version: `npm version patch`, `minor`, or `major`.
|
|
139
|
+
5. Push commit and tag: `git push --follow-tags`.
|
|
140
|
+
6. Check GitHub Actions publish job.
|
|
141
|
+
7. Smoke test after publish: `npx <package-name> --help` or `npx <package-name> list`.
|
|
142
|
+
|
|
143
|
+
## Failure Handling
|
|
144
|
+
|
|
145
|
+
| Symptom | Root Cause | Response |
|
|
146
|
+
| --- | --- | --- |
|
|
147
|
+
| `E403 Two-factor authentication... required` | Manual publish without required 2FA or publish token | Enable npm 2FA and retry manual first publish, or use Trusted Publishing after package exists |
|
|
148
|
+
| `ENEEDAUTH` in GitHub Actions | Trusted Publisher mismatch or missing OIDC permission | Check owner, repo, workflow filename, environment name, and `id-token: write` |
|
|
149
|
+
| npm auto-corrects `package.json` | Metadata is publishable but not canonical | Run `npm pkg fix` or make the exact minimal metadata change |
|
|
150
|
+
| `npm ci` fails in CI | Missing or stale lockfile | Generate/update `package-lock.json` with the intended npm version |
|
|
151
|
+
| Package name unavailable | Name already exists or scope mismatch | Rename to a scoped package such as `@owner/name` and set public access |
|
|
152
|
+
|
|
153
|
+
## Guardrails
|
|
154
|
+
|
|
155
|
+
- Never publish without reading the `npm pack --dry-run` file list.
|
|
156
|
+
- Never add `NPM_TOKEN` when Trusted Publishing works for the project.
|
|
157
|
+
- Never reuse a version after a failed or partial publish; check `npm info <name> versions`.
|
|
158
|
+
- Never claim a package is published until npm or `npx` confirms the released version.
|