pkuppens/issue-gh-bulk-scratch

GitHub bulk issues + local 1-1 scratch files

When to use

• You are creating one umbrella EPIC and several child issues with gh issue create and want local markdown copies that stay in sync (filename = GitHub issue number).
• You need self-assignment so every issue (EPIC and children) appears on the assignee’s list.
• You are renumbering after creation when draft files were ISSUE-1 … ISSUE-n (sequence) before you knew the real numbers.

Prerequisites

• gh auth login (user who should own issues, so @me matches intended assignee)
• tmp/ in repo .gitignore for scratch (see repo CLAUDE.md)
• issue-metadata: default assignee = trigger / operator via --assignee @me

Way of working

1. Draft bodies in gitignored tmp/ (repo root or tmp/github/ if your repo standardizes that):

   • Option A (unknown numbers): tmp/EPIC-DRAFT.md and tmp/ISSUE-DRAFT-1.md … tmp/ISSUE-DRAFT-N.md — import order = creation order, not the final issue number.
   • Option B (after creation): one file per issue: tmp/EPIC-<#>.md, tmp/ISSUE-<#>.md (GitHub number in the name).

2. Create on GitHub (PowerShell: parse number from the URL line https://github.com/OWNER/REPO/issues/N if gh issue create has no --json number in your gh version):

   • EPIC first; optional gh issue comment on the EPIC listing children.
   • For each child: gh issue create --title "..." --body-file tmp/... --assignee @me (or run Assign in step 4 in one pass).

3. Rename to 1-1 (required): after you know the numbers, write final files as tmp/EPIC-<N>.md and tmp/ISSUE-<N>.md and delete the draft names. The integer in the filename must equal the GitHub issue number. Optional first line in each file:

   > **Local scratch** for GitHub **#N** — [open issue](.../issues/N) · EPIC: [#PARENT](.../issues/PARENT)
   

4. Self-assign (required for visibility on “my” list): for EPIC and every child, either:

   • pass --assignee @me on each gh issue create, or
   • after creation: gh issue edit <n> --add-assignee @me for all N in one loop.

   PowerShell: @me must be quoted or the shell treats @ as splatting — use --add-assignee '@me' (or assign the login from gh api user -q .login).

5. Record mapping in tmp/GITHUB-REFS.md (gitignored) or tmp/GITHUB-REFS.md + table: Kind | # | URL | local file.

6. Link children to EPIC: gh issue comment on the EPIC with a checklist - #16 …, and optionally a short comment on each child: Part of portfolio EPIC #15.

gh CLI (examples)

# Create with assignee
gh issue create -t "Title" -F tmp/EPIC-DRAFT.md --assignee @me

# Self-assign in batch (bash)
for n in 15 16 17 18; do gh issue edit "$n" --add-assignee @me; done

Integration

• Composes with issue-workflow (duplicates, acceptance criteria) and issue-metadata (assignee rules).
• After issues exist, work follows the repo branch strategy (plan-branch-strategy, plan).

Anti-patterns

• Committing tmp/* to git unless the project explicitly version-controls drafts.
• Leaving assignees empty when the human requester is known — use @me or the requester’s login.
• Keeping only ISSUE-1 … ISSUE-10 as final names when GitHub actually issued #16…#25 (breaks 1-1 reference).