create-meta-prompts
Creates optimized prompts for Claude-to-Claude pipelines with research, planning, and execution stages. Activates when building prompts that produce outputs for other prompts to consume, or running mu
SKILL.mdplugins/loxihart-ai/skills/create-meta-prompts/SKILL.md
GitHubCreate Meta-Prompts
Builds structured prompts for multi-stage Claude-to-Claude workflows. Every
execution produces a SUMMARY.md for human scanning. Each prompt gets its own
folder in .prompts/ with clear provenance and chain detection.
Workflow (7 Steps)
- Intake — Determine purpose (Do/Plan/Research/Refine), gather requirements
- Chain detection — Check
.prompts/*/for existing research/plan files to reference - Generate — Create prompt using purpose-specific patterns
- Save — Create folder in
.prompts/{number}-{topic}-{purpose}/ - Present — Show decision tree for running
- Execute — Run prompt(s) with dependency-aware execution engine
- Summarize — Create
SUMMARY.mdfor human scanning
Folder Structure
.prompts/
├── 001-auth-research/
│ ├── completed/001-auth-research.md ← archived after run
│ ├── auth-research.md ← full output (XML for Claude)
│ └── SUMMARY.md ← executive summary (human-readable)
├── 002-auth-plan/
│ ├── completed/002-auth-plan.md
│ ├── auth-plan.md
│ └── SUMMARY.md
└── 003-auth-implement/
├── completed/003-auth-implement.md
└── SUMMARY.md
Purpose Routing
| Purpose | Load pattern file | Use when |
|---|---|---|
| Do | references/do-patterns.md |
Build/implement/fix/create |
| Plan | references/plan-patterns.md |
Roadmap/approach/strategy |
| Research | references/research-patterns.md |
Gather/analyze/explore |
| Refine | references/refine-patterns.md |
Improve existing output |
Intake
See references/intake-gate.md for the full intake flow.
Questions by purpose: references/question-bank.md.
Prompt Structure (All Outputs)
Every generated prompt includes:
- Objective — What to accomplish, why it matters
- Context — Referenced files (
@), dynamic context (!) - Requirements — Specific instructions
- Output specification — Where to save, what structure
- Metadata (research/plan only) —
<confidence>,<dependencies>,<open_questions>,<assumptions> - SUMMARY.md requirement — Mandatory in every prompt
- Success criteria
Execution Engine
See references/execution-engine.md for full details on:
- Single / Sequential / Parallel / Mixed-DAG execution modes
- Dependency detection from
@references - Output validation rules
- Failure handling and recovery options
- Archiving behavior
- Result presentation format
Reference Index
| File | Purpose |
|---|---|
references/intake-gate.md |
Intake flow, purpose detection, chain detection |
references/execution-engine.md |
Full execution engine specification |
references/do-patterns.md |
Do prompt patterns |
references/plan-patterns.md |
Plan prompt patterns |
references/research-patterns.md |
Research prompt patterns |
references/refine-patterns.md |
Refine/iteration patterns |
references/summary-template.md |
SUMMARY.md structure |
references/metadata-guidelines.md |
Confidence/dependencies/assumptions fields |
references/question-bank.md |
Intake questions by purpose |
references/intelligence-rules.md |
Extended thinking, parallel tools, depth |
Success Criteria
Prompt creation: Intake complete → chain detection done → prompt generated with correct structure → folder created → SUMMARY.md requirement included → decision tree presented.
Execution: Dependencies detected and ordered → prompts run in correct mode → output validated → SUMMARY.md displayed inline → failed prompts handled with recovery options → successful prompts archived.