How this skill is triggered — by the user, by Claude, or both
Slash command
/lisa:create-epicThe summary Claude sees in its skill listing — used to decide when to auto-load this skill
**Epic name:** Provided via arguments (already hyphenated)
Epic name: Provided via arguments (already hyphenated)
Check if .lisa/epics/<name>/.state exists.
If exists:
Epic '<name>' already exists.
Use /lisa:epic-status <name> to see status, or /lisa:continue <name> to resume.
STOP.
Create directory .lisa/epics/<name>/ and write .lisa/epics/<name>/.state:
{
"name": "<name>",
"currentPhase": "spec",
"specComplete": false,
"researchComplete": false,
"planComplete": false,
"executeComplete": false,
"lastUpdated": "<ISO timestamp>"
}
Also ensure .lisa/ exists and create .lisa/config.jsonc if it doesn't exist:
{
// Lisa Configuration
"execution": {
"maxRetries": 3,
"maxParallelTasks": 1
},
"git": {
"completionMode": "none",
"branchPrefix": "epic/",
"autoPush": true
},
"yolo": {
"defaultMaxIterations": 100
}
}
Have a natural conversation to cover:
Be conversational. Ask clarifying questions. Push back if scope is too large or vague. Keep concise — aim for 20-50 lines total in the final spec.
Show the spec and ask:
Here's the spec:
[formatted spec content]
Ready to save to .lisa/epics/<name>/spec.md?
On confirmation, save:
# Epic: <name>
## Goal
[What we're building and why - 1-2 sentences]
## Scope
- [Included item]
- [Included item]
### Out of Scope
- [Excluded item]
## Acceptance Criteria
- [ ] [Measurable criterion]
- [ ] [Measurable criterion]
## Technical Constraints
- [Constraints or "None"]
Update .state:
{
"specComplete": true,
"currentPhase": "spec",
"lastUpdated": "<ISO timestamp>"
}
Spec saved to .lisa/epics/<name>/spec.md
Next steps:
• /lisa:continue <name> - Continue to research phase (interactive, with checkpoints)
• /lisa:yolo <name> - Auto-execute all remaining phases
• /lisa:list-epics - See all your epics
DO NOT continue automatically. STOP here and wait for the user.
npx claudepluginhub fractalswift/llisa --plugin lisaCreates epic JSON files with vision (overview, goals, metrics, scope, NFRs) via user dialog, ID generation, and existence checks. Deprecated; use /plan instead.
Guides spec-driven development through phases: research, requirements, design, tasks, autonomous task-by-task implementation. Decomposes large epics into ordered specs with progress tracking.
Guides creation, editing, and verification of skills for AI coding agents using test-driven development with subagent scenarios. Use when authoring or debugging skills.