create-storylisted
Install: claude install-skill rterakedis/leanwheel-skills
# Create Story Skill
**Goal:** Story file complete enough that dev session needs only the story. All decisions, patterns, constraints in Dev Notes — thin stories cause rework.
## Activation
### Step 1 — Identify target story
- If user specified epic/story (e.g., "1-2"), use it.
- Otherwise read `docs/epics.md`, find first story with no file in `docs/epics/`. Scan only epics with live `### Story` bodies — a collapsed epic (summary table) is closed; its rows already have files and are pointers, not specs.
- Confirm: "Creating story {epic}-{story}: {title}?"
### Step 2 — Load documents via cache
Check cache validity:
```bash
[ -f docs/epics/epic-{epic_num}-context.md ] \
&& [ docs/epics/epic-{epic_num}-context.md -nt docs/prd.md ] \
&& [ docs/epics/epic-{epic_num}-context.md -nt docs/architecture.md ] \
&& echo "valid" || echo "stale"
```
**If valid:**
- Read `docs/epics/epic-{epic_num}-context.md` (replaces prd + architecture)
- Read relevant story entry in `docs/epics.md`
- Read completed `docs/epics/{epic_num}-*.md` for learnings
- Tell user: "Using epic context cache."
**If missing/stale:**
- Read `docs/prd.md`, `docs/architecture.md`, `docs/epics.md` fully
- Read completed `docs/epics/{epic_num}-*.md`
- Generate cache before writing story
- Tell user: "Generated epic-{epic_num} context cache."
**Collapsed epics (either branch):** the spec for a summary-table row lives in the linked `docs/epics/{N}-{M}-*.md` — read that file for prior-story detail, not `docs/ep