github-create-issues

Compare original and translation side by side

🇺🇸

Original

English

🇨🇳

Translation

Chinese

github-create-issues

Take a

github_backlog_input.json

and create the issues in GitHub by driving the bundled

create_github_issues.py

. The script creates the milestone and labels if needed, creates each issue, then creates a tracking issue with a task list linking them all.

Creating issues is effectively un-undoable (close is manual; notifications already fired). So this skill is built around a hard gate: visual dry-run always, real run only on an explicit "yes".

For the exact JSON shapes see

${CLAUDE_PLUGIN_ROOT}/references/data-contracts.md

通过运行内置的

create_github_issues.py

脚本，读取

github_backlog_input.json

文件并在GitHub中创建Issues。该脚本会在需要时创建里程碑和标签，逐个创建Issue，然后生成一个包含所有Issue任务列表的追踪Issue。

创建Issue的操作基本无法撤销（关闭需手动操作，且通知已发出）。因此该技能设置了严格的确认机制：始终先执行可视化预执行，仅在用户明确回复“yes”后才执行实际创建操作。

关于具体的JSON格式，请参考

${CLAUDE_PLUGIN_ROOT}/references/data-contracts.md

。

Prerequisites

前置条件

Auth —
```
gh auth login
```
once (the script reads
```
gh auth token
```
). Or set
```
$env:GH_TOKEN
```
. See the
```
github-auth
```
skill.
Owner/repo — set
```
$env:GH_OWNER
```
/
```
$env:GH_REPO
```
, or include
```
owner
```
/
```
repo
```
in the JSON. Env vars override the JSON.

Python + requests + openpyxl — run the setup checker if unsure:

powershell -ExecutionPolicy Bypass -File "${CLAUDE_PLUGIN_ROOT}/scripts/setup_check_github.ps1"

身份验证 — 先执行
```
gh auth login
```
（脚本会读取
```
gh auth token
```
）。或者设置环境变量
```
$env:GH_TOKEN
```
。可查看
```
github-auth
```
技能了解详情。
仓库所有者/仓库名 — 设置环境变量
```
$env:GH_OWNER
```
/
```
$env:GH_REPO
```
，或在JSON文件中包含
```
owner
```
/
```
repo
```
字段。环境变量优先级高于JSON文件中的配置。

Python + requests + openpyxl — 若不确定是否已安装，可运行检查脚本：

powershell -ExecutionPolicy Bypass -File "${CLAUDE_PLUGIN_ROOT}/scripts/setup_check_github.ps1"

Step 1 — VISUAL DRY-RUN (creates nothing)

步骤1 — 可视化预执行（不会创建任何内容）

GitHub Issues has no

validateOnly

API. Instead: read

github_backlog_input.json

and render a formatted table for the user to review before any API call.

Present:

 # | Key    | Title                              | Labels               | Assignee
---|--------|------------------------------------|----------------------|---------
 1 | row-1  | Portal label "Auto" should show... | bug, P1, size:S      | —
 2 | row-2  | Add rate limiting to API           | enhancement, P2, ... | pon
...
Milestone: Audit Wave 1
Tracking issue: yes (created last, links all above)
Total: 2 issues

GATE — stop here. Do not proceed until the user has explicitly approved the list (e.g. "create them", "go ahead", "yes").

If the user requests changes, update

github_backlog_input.json

and re-render the table. The table is free — re-rendering costs nothing.

GitHub Issues没有

validateOnly

API。因此我们采取的方式是：读取

github_backlog_input.json

文件，渲染格式化表格供用户在调用任何API前进行审核。

展示格式如下：

 # | Key    | Title                              | Labels               | Assignee
---|--------|------------------------------------|----------------------|---------
 1 | row-1  | Portal label "Auto" should show... | bug, P1, size:S      | —
 2 | row-2  | Add rate limiting to API           | enhancement, P2, ... | pon
...
Milestone: Audit Wave 1
Tracking issue: yes (created last, links all above)
Total: 2 issues

确认关卡 — 在此暂停。在用户明确批准该列表前（例如“创建它们”“继续”“yes”），请勿继续执行。

如果用户要求修改，请更新

github_backlog_input.json

并重新渲染表格。渲染表格无任何成本，可以反复操作。

Step 2 — REAL RUN (only after explicit approval)

步骤2 — 实际执行（仅在明确批准后进行）

powershell

$env:GH_OWNER = "Cartagena365"
$env:GH_REPO  = "GlassHull"
python "${CLAUDE_PLUGIN_ROOT}/scripts/create_github_issues.py" `
  --input  "<workdir>/github_backlog_input.json" `
  --output "<workdir>/github_backlog_result.json"

The script prints one line per created issue (

key X -> #N <url>

) and a final

tracking issue -> #N <url>

. Then it writes

github_backlog_result.json

Relay the output to the user. Surface any failure clearly — a partial run is better visible than silent.

powershell

$env:GH_OWNER = "Cartagena365"
$env:GH_REPO  = "GlassHull"
python "${CLAUDE_PLUGIN_ROOT}/scripts/create_github_issues.py" `
  --input  "<workdir>/github_backlog_input.json" `
  --output "<workdir>/github_backlog_result.json"

脚本会为每个创建的Issue打印一行信息（格式为

key X -> #N <url>

），最后打印追踪Issue的信息（

tracking issue -> #N <url>

）。随后会生成

github_backlog_result.json

文件。

将脚本输出信息同步给用户。若出现失败需清晰告知用户——部分执行失败比静默失败更易被察觉。

After creating — verify

创建完成后 — 验证

Confirm the items landed. Check

github_backlog_result.json

— every item should have a

number

and

url

. Items without a

number

failed; surface them to the user.

确认所有条目已成功创建。检查

github_backlog_result.json

文件——每个条目都应包含

number

和

url

字段。没有

number

字段的条目表示创建失败，需告知用户。

Idempotency — don't double-create

幂等性 — 避免重复创建

create_github_issues.py

is not idempotent: re-running on the same

github_backlog_input.json

creates a second set of issues. After a successful run, hand off to github-writeback-tracking to stamp

Issue #

Issue URL

back onto source rows. Once a row carries an Issue #, treat it as done. To re-file a subset, prune the input to only the un-filed

key

s first.

create_github_issues.py

不具备幂等性：对同一个

github_backlog_input.json

文件重复运行脚本会创建第二组相同的Issues。执行成功后，将任务移交至github-writeback-tracking，将

Issue #

Issue URL

标记回源数据行。一旦某行已标记Issue编号，即视为已完成。若需重新提交部分条目，请先将输入文件中已提交的

key

条目移除。