@weotro/dx 0.1.5 → 0.1.7
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/README.md +2 -0
- package/agent-references/ship-issue-pr-core.md +502 -0
- package/lib/codex-initial.js +64 -0
- package/package.json +2 -1
- package/skills/cc-ship-issue-pr/SKILL.md +27 -0
- package/skills/cc-ship-issue-pr/agents/openai.yaml +2 -0
- package/skills/cc-ship-issue-pr/references/runtime.md +96 -0
- package/skills/git-release/SKILL.md +18 -3
- package/skills/git-release/agents/openai.yaml +2 -2
- package/skills/git-release/references/post-release-follow-up.md +176 -0
- package/skills/oo-ship-issue-pr/SKILL.md +27 -0
- package/skills/oo-ship-issue-pr/agents/openai.yaml +2 -0
- package/skills/oo-ship-issue-pr/references/runtime.md +92 -0
- package/skills/prune-git-repository/SKILL.md +83 -0
- package/skills/prune-git-repository/agents/openai.yaml +6 -0
- package/skills/prune-git-repository/references/pitfalls.md +43 -0
- package/skills/prune-git-repository/scripts/prune-branches.sh +136 -0
- package/skills/prune-git-repository/scripts/prune-tags-releases.sh +117 -0
- package/skills/prune-git-repository/scripts/prune-worktrees.sh +71 -0
- package/skills/create-issue/SKILL.md +0 -90
- package/skills/ship-issue-pr/SKILL.md +0 -742
- package/skills/third-party-review/SKILL.md +0 -64
- package/skills/third-party-review/agents/openai.yaml +0 -6
|
@@ -1,64 +0,0 @@
|
|
|
1
|
-
---
|
|
2
|
-
name: third-party-review
|
|
3
|
-
description: ""
|
|
4
|
-
---
|
|
5
|
-
|
|
6
|
-
# Third-Party Review
|
|
7
|
-
|
|
8
|
-
只执行已经批准的审查任务。调用门槛、简单修改判断、次数预算和业务决策均由调用方负责。
|
|
9
|
-
|
|
10
|
-
仅通过以下方式启动:
|
|
11
|
-
|
|
12
|
-
- 用户明确调用 `$third-party-review`;
|
|
13
|
-
- 另一个已激活技能明确写出并调用 `$third-party-review`。
|
|
14
|
-
|
|
15
|
-
不要根据“代码审查”“重大决策”等普通关键词自行触发。
|
|
16
|
-
|
|
17
|
-
## 上下文补全
|
|
18
|
-
|
|
19
|
-
不要要求调用方提供固定参数。启动后先从当前对话、调用技能、仓库状态、Issue/PR 和已有工具结果中推断:
|
|
20
|
-
|
|
21
|
-
- 审查模式:代码审查或决策审查;
|
|
22
|
-
- 审查对象与范围;
|
|
23
|
-
- 调用理由和上游已经做出的初步判断;
|
|
24
|
-
- 完成审查所需的 diff、源码、验证结果或决策材料。
|
|
25
|
-
|
|
26
|
-
优先自行读取可安全获取的信息。只有关键内容无法从上下文或仓库获得时才反问;一次集中询问所有真正缺失的信息,避免逐项追问。
|
|
27
|
-
|
|
28
|
-
反问与用户补充属于当前 `$third-party-review` 调用的延续,不得重新调用技能,也不得重复计算调用次数。收到补充后从暂停点继续执行。
|
|
29
|
-
|
|
30
|
-
若能确定审查对象但部分辅助信息缺失,标注假设并继续。只有连审查对象或目标都无法确定时才必须反问。
|
|
31
|
-
|
|
32
|
-
## Reviewer 路由
|
|
33
|
-
|
|
34
|
-
派发前一次性选择 reviewer:
|
|
35
|
-
|
|
36
|
-
| 当前宿主 | reviewer |
|
|
37
|
-
|---|---|
|
|
38
|
-
| Codex | 已暴露且可直接调用 Claude Code 时使用 Claude Code;否则派一个 fresh-context reviewer subagent |
|
|
39
|
-
| Claude Code | 有匹配的 `codex:*` 技能时使用该技能;代码审查优先 `codex:rescue`,否则派一个 fresh-context reviewer subagent |
|
|
40
|
-
| 其他宿主 | 派一个 fresh-context reviewer subagent |
|
|
41
|
-
|
|
42
|
-
一次任务只派发一个 reviewer。调用失败或超时后不重试、不改派第二个 reviewer;返回 `UNAVAILABLE`。reviewer 不得递归委派。
|
|
43
|
-
|
|
44
|
-
## 执行流程
|
|
45
|
-
|
|
46
|
-
1. 从上下文补全审查任务;必要时集中反问并等待补充。
|
|
47
|
-
2. 不重新评估调用理由是否达到上游技能的门槛。
|
|
48
|
-
3. 移除 token、密码、私钥、真实环境变量与无关个人数据。
|
|
49
|
-
4. 选择一个 reviewer,要求只读:不改文件、不执行 Git/GitHub 写操作、不发布评论、不提交、不推送、不合并。
|
|
50
|
-
5. 传递必要原始材料和调用方的初步结论,要求 reviewer 主动寻找反例、遗漏、错误假设与回归风险。
|
|
51
|
-
6. 返回 reviewer 原始结论和结构化摘要,不替调用方接受、拒绝或落实 findings。
|
|
52
|
-
|
|
53
|
-
## 审查要求
|
|
54
|
-
|
|
55
|
-
`code-review` 要求每条 finding 包含:严重级、稳定位置、证据、影响和最小安全修复。无 finding 时说明检查范围和剩余风险。
|
|
56
|
-
|
|
57
|
-
`decision-review` 要求列出:最强反对理由、隐藏假设、替代方案、兼容性或失败风险、需要补充的证据,以及是否建议维持调用方的初步选择。
|
|
58
|
-
|
|
59
|
-
## 输出契约
|
|
60
|
-
|
|
61
|
-
- `REVIEWED`:reviewer、审查范围、原始结论、结构化 findings 和剩余风险。
|
|
62
|
-
- `UNAVAILABLE`:已选 reviewer、失败或超时原因;明确没有重试或改派。
|
|
63
|
-
|
|
64
|
-
第三方意见仅作为证据。调用方负责事实核验、采纳/拒绝理由和最终结论。
|
|
@@ -1,6 +0,0 @@
|
|
|
1
|
-
interface:
|
|
2
|
-
display_name: "Third-Party Review"
|
|
3
|
-
short_description: "Execute one approved independent adversarial review"
|
|
4
|
-
default_prompt: "Use $third-party-review to execute one approved read-only adversarial review, inferring context and asking once if essential information is missing."
|
|
5
|
-
policy:
|
|
6
|
-
allow_implicit_invocation: false
|