// Use when breaking a plan, spec, or PRD into local markdown task files as vertical-slice tracer bullets. No GitHub. Each slice gets its own file with status/blocked-by/type frontmatter.
[HINT] Download the complete skill directory including SKILL.md and all related files
| name | issues-create-local |
| description | Use when breaking a plan, spec, or PRD into local markdown task files as vertical-slice tracer bullets. No GitHub. Each slice gets its own file with status/blocked-by/type frontmatter. |
Break a plan into local markdown task files using vertical slices (tracer bullets). Same vertical-slice discipline as issues-create, but writes files to plans/issues/<prd>/ instead of calling gh issue create.
issues-createissues-create — collaborative work where slices need to be grabbed from a backlog by humans/bots via GitHub.Work from whatever is already in the conversation. If the user passes a PRD path or name (e.g. auth-system), read plans/auth-system.md. If there's exactly one PRD in plans/, use it. If none, tell the user to run /prd-create first.
Derive <prd> — the kebab-case PRD basename without .md (e.g. plans/auth-system.md → auth-system). All issue files for this PRD go under plans/issues/<prd>/.
CONTEXT.md runs automatically when the skill loads — output replaces the line below:
cat CONTEXT.md 2>/dev/null || trueIf content is present, reuse its vocabulary in slice titles and descriptions. If you have not already explored the codebase, do so now to ground the slices in real modules.
Break the plan into tracer bullet issues. Each issue is a thin vertical slice that cuts through ALL integration layers end-to-end, NOT a horizontal slice of one layer.
Slices may be HITL or AFK:
Prefer AFK over HITL where possible.
Vertical slice rules:
Present the proposed breakdown as a numbered list. For each slice, show:
Ask the user:
Iterate until the user approves the breakdown.
For each approved slice, write one file to plans/issues/<prd>/NN-<slug>.md.
NN — two-digit zero-padded ordinal in dependency order (blockers first): 01, 02, …<slug> — kebab-case version of the title (lowercase, hyphens, no punctuation), e.g. add-login-formmkdir -p plans/issues/<prd>
Use the file template below. Reference blockers by their filename basename (e.g. 01-add-login-form) — that survives renames inside this directory better than absolute paths.
---
title: <Short descriptive title>
status: open
type: AFK # or HITL
blocked-by: [] # or [01-add-login-form, 02-add-session-store]
user-stories: [] # or [1, 4, 7] — story numbers from the PRD
---
## What to build
A concise description of this vertical slice. Describe the end-to-end behavior, not layer-by-layer implementation.
## Acceptance criteria
- [ ] Criterion 1
- [ ] Criterion 2
- [ ] Criterion 3
## Notes
Any context, references, or open questions for the implementer.
Print the list of files written, with paths relative to the repo root, e.g.:
Wrote 5 slices to plans/issues/auth-system/:
01-add-login-form.md (AFK)
02-add-session-store.md (AFK, blocked-by: 01)
03-protect-dashboard-route.md (AFK, blocked-by: 02)
04-logout-button.md (AFK, blocked-by: 02)
05-session-expiry-design.md (HITL)
Then point the user at /ralph-loop <prd> to start working through them. ralph-loop reads these files as its task queue: it picks the lowest-numbered open slice whose blockers are all status: done, uses the slice's acceptance criteria as the task spec, flips checkboxes commit-by-commit (strict — only criteria fulfilled by that commit), and flips status: open → done automatically when the last - [ ] becomes - [x]. The overseer audits each flip against the diff and rolls back overclaims.
ralph-loop iterations maintain status and acceptance-criteria checkboxes automatically under the strict checkbox protocol — no manual flips needed during AFK runs. The overseer reconciles overclaims and stale checkboxes on its tick. To list remaining work at a glance:
grep -L "status: done" plans/issues/<prd>/*.md