[{"data":1,"prerenderedAt":-1},["ShallowReactive",2],{"skill:01M6XGC2AYQYGWS41A2ZKBDHXH":3,"skill-same:01M6XGC2AYQYGWS41A2ZKBDHXH":76},{"skillNo":4,"name":5,"nameZh":6,"description":7,"descriptionZh":6,"repositoryFullName":8,"ownerLogin":9,"ownerAvatarUrl":10,"ownerAccountNo":11,"ownerDisplayName":12,"repositoryNo":13,"starsCount":14,"licenseSpdxId":15,"licenseFileUrl":6,"sourceHtmlUrl":16,"categories":17,"keywords":18,"publishedAt":39,"linkOnly":40,"markdownContent":41,"markdownBody":42,"frontMatter":43,"refName":56,"commitSha":57,"skillDirectory":58,"forksCount":14,"repositoryHtmlUrl":59,"repositoryHomepageUrl":6,"ownerHtmlUrl":60,"ownerBlogUrl":6,"ownerDescription":6,"updatedAt":61,"install":62,"resources":75},"01M6XGC2AYQYGWS41A2ZKBDHXH","model-economics",null,"Cost modeling and ROI analysis for specialized LLM development. Use when deciding whether to train a custom model, estimating total cost, or calculating break-even vs frontier APIs. Covers training costs, inference costs, and time-to-ROI projections.","fadrienne\u002FMyGitHub","fadrienne","https:\u002F\u002Favatars.githubusercontent.com\u002Fu\u002F197874571?v=4","01M6X64V8VWCRSMFZ4EPS2BWXG","Francesca Adrienne","01M6X64VD5MW4F9PHDWQKS7ZAF",0,"MIT","https:\u002F\u002Fgithub.com\u002Ffadrienne\u002FMyGitHub\u002Fblob\u002Fmain\u002Fopenscience\u002Fbackend\u002Fcli\u002Fskills\u002Fml-training\u002Fmodel-economics\u002FSKILL.md",[],[19,20,21,22,23,24,25,26,27,28,29,30,31,32,33,34,35,36,37,38],"model economics","model","economics","synthetic","Training Cost","when","flywheel","break","for","roi analysis","projections","roi","modeling","time","business case","specialized","analysis","development","training","cost analysis","2026-11-10 10:12:30",false,"---\nname: model-economics\ndescription: Cost modeling and ROI analysis for specialized LLM development. Use when deciding whether to train a custom model, estimating total cost, or calculating break-even vs frontier APIs. Covers training costs, inference costs, and time-to-ROI projections.\ncategory: ml-training\nversion: 1.0.0\nauthor: Synthetic Sciences\nlicense: MIT\ntags: [Economics, Cost Analysis, ROI, Break-Even, Training Cost, Inference Cost, Model Flywheel, Business Case]\ndependencies: []\n---\n\n# Model Economics: Cost Analysis for LLM Specialization\n\n## When to Use This Skill\n\nUse this skill as the **first step** of any model-specialization assessment to determine:\n- Is it worth training a specialized model?\n- What will the total investment be?\n- When will the specialized model break even vs frontier APIs?\n- What's the ongoing cost comparison?\n\n## Frontier API Cost Model\n\n**Important**: Use `websearch` to get current pricing — these change frequently. Reference tiers as of early 2026:\n\n### Premium Tier ($5-25\u002FM output tokens)\n| Model | Input ($\u002FM) | Output ($\u002FM) | Notes |\n|-------|-------------|--------------|-------|\n| Claude Opus 4.6 | $5.00 | $25.00 | Highest capability |\n| GPT-5.2 Pro | $21.00 | $168.00 | Reasoning-heavy |\n\n**Savings potential**: Highest. A specialized 8B model costs ~$0.05-0.15\u002FM tokens to serve = 100-500x cheaper.\n\n### Standard Tier ($1.75-14\u002FM output tokens)\n| Model | Input ($\u002FM) | Output ($\u002FM) | Notes |\n|-------|-------------|--------------|-------|\n| GPT-5.2 Chat | $1.75 | $14.00 | Most popular |\n| Claude Sonnet 4.5 | $3.00 | $15.00 | Good balance |\n| Qwen3 Max | $1.20 | $6.00 | Cost-effective |\n\n**Savings potential**: Strong at volume. Break-even is achievable with moderate traffic.\n\n### Budget Tier ($0.10-3\u002FM output tokens)\n| Model | Input ($\u002FM) | Output ($\u002FM) | Notes |\n|-------|-------------|--------------|-------|\n| Gemini 3 Flash | $0.50 | $3.00 | Fast, cheap |\n| GLM 4.7 | $0.40 | $1.50 | Very cost-effective |\n| Mistral Small | $0.10 | $0.30 | Cheapest commercial |\n\n**Savings potential**: Harder to beat on cost alone. Specialize for quality, not cost.\n\n### Free \u002F Near-Free\nMany open-weight models available on OpenRouter at $0. If you only need basic capability, self-hosting these is essentially free beyond GPU cost.\n\n## Training Cost Estimation\n\n### Tinker (Managed LoRA)\nCheapest option for supported models. No GPU management.\n\n| Model Size | Method | Cost\u002F1K examples | Duration |\n|------------|--------|-------------------|----------|\n| 8B | LoRA | ~$2-5 | 15-30 min |\n| 14B | LoRA | ~$5-10 | 30-60 min |\n| 32B | LoRA | ~$10-25 | 1-3 hours |\n| 70B | LoRA | ~$25-60 | 3-8 hours |\n\nLoad `tinker-training-cost` skill for exact per-token rates.\n\n### Modal (Serverless GPU)\n\n| GPU | Cost\u002FHour | Best For |\n|-----|-----------|----------|\n| A100-40GB | ~$1.10 | 7-14B LoRA, 7B full |\n| A100-80GB | ~$1.60 | 14-32B LoRA, 14B full |\n| H100-80GB | ~$3.25 | 32-70B LoRA, large batches |\n\n**Typical training runs**:\n- 8B LoRA, 5K examples, 3 epochs: ~1-2 hours on A100 = $1-3\n- 14B LoRA, 10K examples, 3 epochs: ~3-5 hours on A100-80GB = $5-8\n- 70B LoRA, 10K examples, 2 epochs: ~8-12 hours on H100 = $26-39\n\n### TensorPool (Dedicated GPU)\n\n| GPU | Cost\u002FHour | Best For |\n|-----|-----------|----------|\n| A100-80GB | ~$1.50 | Standard training |\n| H200-141GB | ~$3.00 | Large models, fast training |\n| B200-192GB | ~$4.50 | Cutting-edge, maximum throughput |\n\nBest for: Multi-day training, large-scale RL, dedicated inference.\n\n## Inference Cost Comparison\n\nThe key calculation: what does it cost to serve the specialized model vs the frontier API?\n\n### Self-Hosted Inference Cost\n\n```python\ndef inference_cost_per_million_tokens(gpu_cost_per_hour, tokens_per_second):\n    \"\"\"Calculate cost per million tokens for self-hosted inference.\n\n    Args:\n        gpu_cost_per_hour: GPU rental cost (e.g., $3.25 for H100)\n        tokens_per_second: Model throughput (varies by model size and batch)\n    \"\"\"\n    tokens_per_hour = tokens_per_second * 3600\n    cost_per_token = gpu_cost_per_hour \u002F tokens_per_hour\n    cost_per_million = cost_per_token * 1_000_000\n    return cost_per_million\n```\n\n### Reference Throughput (vLLM, single GPU)\n\n| Model Size | GPU | Tokens\u002Fsec (batch) | Cost\u002FM tokens |\n|------------|-----|-------------------|---------------|\n| 8B | A100-80GB | ~3,000-5,000 | $0.09-0.15 |\n| 8B | H100-80GB | ~5,000-8,000 | $0.11-0.18 |\n| 14B | A100-80GB | ~1,500-3,000 | $0.15-0.30 |\n| 14B | H100-80GB | ~3,000-5,000 | $0.18-0.30 |\n| 32B | H100-80GB | ~800-1,500 | $0.60-1.10 |\n| 70B | 2xH100 | ~500-1,000 | $1.80-3.60 |\n\n**Key insight**: A specialized 8B model on Modal\u002FTensorPool costs **$0.05-0.15\u002FM tokens** vs **$5-25\u002FM** for premium frontier = **50-500x cheaper per request**.\n\n## Break-Even Calculator\n\n```python\ndef break_even_analysis(\n    monthly_api_spend: float,\n    training_cost: float,\n    monthly_inference_cost: float,\n    setup_hours: float = 40,\n    hourly_rate: float = 100,\n):\n    \"\"\"Calculate break-even timeline for model specialization.\n\n    Args:\n        monthly_api_spend: Current monthly frontier API cost\n        training_cost: One-time training cost (GPU + data prep)\n        monthly_inference_cost: Monthly cost to serve specialized model\n        setup_hours: Engineering time to set up pipeline\n        hourly_rate: Engineering cost per hour\n    \"\"\"\n    engineering_cost = setup_hours * hourly_rate\n    total_upfront = training_cost + engineering_cost\n    monthly_savings = monthly_api_spend - monthly_inference_cost\n\n    if monthly_savings \u003C= 0:\n        return {\n            \"viable\": False,\n            \"reason\": \"Specialized model costs more to serve than frontier API\",\n            \"monthly_savings\": monthly_savings,\n        }\n\n    months_to_break_even = total_upfront \u002F monthly_savings\n\n    return {\n        \"viable\": True,\n        \"total_upfront_cost\": total_upfront,\n        \"training_cost\": training_cost,\n        \"engineering_cost\": engineering_cost,\n        \"monthly_api_spend\": monthly_api_spend,\n        \"monthly_inference_cost\": monthly_inference_cost,\n        \"monthly_savings\": monthly_savings,\n        \"months_to_break_even\": round(months_to_break_even, 1),\n        \"annual_savings\": monthly_savings * 12 - total_upfront,\n        \"year_2_savings\": monthly_savings * 12,  # Fully amortized\n    }\n\n\n# Example: Replace GPT-4o for a specific task\nresult = break_even_analysis(\n    monthly_api_spend=5000,       # $5K\u002Fmonth on GPT-4o\n    training_cost=500,            # $500 for LoRA fine-tune + data prep\n    monthly_inference_cost=200,   # $200\u002Fmonth on Modal (8B model)\n    setup_hours=40,               # 1 week of engineering\n    hourly_rate=100,              # $100\u002Fhr engineer\n)\n# Result: Break-even in ~1.1 months, $52K annual savings after year 1\n```\n\n## ROI Timeline Template\n\n| Month | Frontier Cost | Specialized Cost | Cumulative Savings | Notes |\n|-------|--------------|-----------------|-------------------|-------|\n| 0 | — | $4,500 upfront | -$4,500 | Training + engineering |\n| 1 | $5,000 | $200 | $300 | First month live |\n| 2 | $5,000 | $200 | $5,100 | Break-even |\n| 3 | $5,000 | $200 | $9,900 | |\n| 6 | $5,000 | $200 | $24,300 | |\n| 12 | $5,000 | $300 | $52,200 | Includes retrain cost |\n\n## Decision Framework\n\n### Strong Case for Specialization\n- Monthly API spend > $1,000\n- Constrained task (classification, extraction, formatting, domain Q&A)\n- Production data available (logs, corrections, accept\u002Freject signals)\n- Quality requirements are well-defined and measurable\n- Task doesn't change frequently\n\n### Weak Case for Specialization\n- Monthly API spend \u003C $500\n- Diverse, open-ended tasks (general assistant)\n- No production data to train on\n- Rapid model improvement expected (new frontier models releasing soon)\n- Task requires reasoning at the frontier level\n- Small team with no ML engineering capacity\n\n### Hybrid Approach\nRoute easy requests to specialized model, hard ones to frontier:\n- **90% of requests** → Specialized 8B (cheap, fast)\n- **10% of requests** → Frontier fallback (complex, novel)\n- **Result**: 80-90% cost reduction while maintaining quality on hard cases\n\n## Quick Assessment Questions\n\nAsk the user these to determine viability:\n\n1. **What's your monthly frontier API spend?**\n   - \u003C $500: Probably not worth it (low ROI)\n   - $500-2K: Worth investigating\n   - $2K+: Strong candidate\n   - $10K+: Almost certainly worth it\n\n2. **How constrained is the task?**\n   - Single task (classification, extraction): Great fit\n   - Few related tasks: Good fit\n   - Many diverse tasks: Harder to specialize\n\n3. **What production data do you have?**\n   - API logs with user feedback: Excellent\n   - API logs without feedback: Good (can distill)\n   - No production data: Need synthetic bootstrap\n\n4. **What's your quality bar?**\n   - Must match frontier exactly: Harder, may need bigger model\n   - 90% of frontier quality is fine: 8B LoRA likely sufficient\n   - Task-specific metrics (accuracy, format): Easiest to optimize\n","# Model Economics: Cost Analysis for LLM Specialization\n\n## When to Use This Skill\n\nUse this skill as the **first step** of any model-specialization assessment to determine:\n- Is it worth training a specialized model?\n- What will the total investment be?\n- When will the specialized model break even vs frontier APIs?\n- What's the ongoing cost comparison?\n\n## Frontier API Cost Model\n\n**Important**: Use `websearch` to get current pricing — these change frequently. Reference tiers as of early 2026:\n\n### Premium Tier ($5-25\u002FM output tokens)\n| Model | Input ($\u002FM) | Output ($\u002FM) | Notes |\n|-------|-------------|--------------|-------|\n| Claude Opus 4.6 | $5.00 | $25.00 | Highest capability |\n| GPT-5.2 Pro | $21.00 | $168.00 | Reasoning-heavy |\n\n**Savings potential**: Highest. A specialized 8B model costs ~$0.05-0.15\u002FM tokens to serve = 100-500x cheaper.\n\n### Standard Tier ($1.75-14\u002FM output tokens)\n| Model | Input ($\u002FM) | Output ($\u002FM) | Notes |\n|-------|-------------|--------------|-------|\n| GPT-5.2 Chat | $1.75 | $14.00 | Most popular |\n| Claude Sonnet 4.5 | $3.00 | $15.00 | Good balance |\n| Qwen3 Max | $1.20 | $6.00 | Cost-effective |\n\n**Savings potential**: Strong at volume. Break-even is achievable with moderate traffic.\n\n### Budget Tier ($0.10-3\u002FM output tokens)\n| Model | Input ($\u002FM) | Output ($\u002FM) | Notes |\n|-------|-------------|--------------|-------|\n| Gemini 3 Flash | $0.50 | $3.00 | Fast, cheap |\n| GLM 4.7 | $0.40 | $1.50 | Very cost-effective |\n| Mistral Small | $0.10 | $0.30 | Cheapest commercial |\n\n**Savings potential**: Harder to beat on cost alone. Specialize for quality, not cost.\n\n### Free \u002F Near-Free\nMany open-weight models available on OpenRouter at $0. If you only need basic capability, self-hosting these is essentially free beyond GPU cost.\n\n## Training Cost Estimation\n\n### Tinker (Managed LoRA)\nCheapest option for supported models. No GPU management.\n\n| Model Size | Method | Cost\u002F1K examples | Duration |\n|------------|--------|-------------------|----------|\n| 8B | LoRA | ~$2-5 | 15-30 min |\n| 14B | LoRA | ~$5-10 | 30-60 min |\n| 32B | LoRA | ~$10-25 | 1-3 hours |\n| 70B | LoRA | ~$25-60 | 3-8 hours |\n\nLoad `tinker-training-cost` skill for exact per-token rates.\n\n### Modal (Serverless GPU)\n\n| GPU | Cost\u002FHour | Best For |\n|-----|-----------|----------|\n| A100-40GB | ~$1.10 | 7-14B LoRA, 7B full |\n| A100-80GB | ~$1.60 | 14-32B LoRA, 14B full |\n| H100-80GB | ~$3.25 | 32-70B LoRA, large batches |\n\n**Typical training runs**:\n- 8B LoRA, 5K examples, 3 epochs: ~1-2 hours on A100 = $1-3\n- 14B LoRA, 10K examples, 3 epochs: ~3-5 hours on A100-80GB = $5-8\n- 70B LoRA, 10K examples, 2 epochs: ~8-12 hours on H100 = $26-39\n\n### TensorPool (Dedicated GPU)\n\n| GPU | Cost\u002FHour | Best For |\n|-----|-----------|----------|\n| A100-80GB | ~$1.50 | Standard training |\n| H200-141GB | ~$3.00 | Large models, fast training |\n| B200-192GB | ~$4.50 | Cutting-edge, maximum throughput |\n\nBest for: Multi-day training, large-scale RL, dedicated inference.\n\n## Inference Cost Comparison\n\nThe key calculation: what does it cost to serve the specialized model vs the frontier API?\n\n### Self-Hosted Inference Cost\n\n```python\ndef inference_cost_per_million_tokens(gpu_cost_per_hour, tokens_per_second):\n    \"\"\"Calculate cost per million tokens for self-hosted inference.\n\n    Args:\n        gpu_cost_per_hour: GPU rental cost (e.g., $3.25 for H100)\n        tokens_per_second: Model throughput (varies by model size and batch)\n    \"\"\"\n    tokens_per_hour = tokens_per_second * 3600\n    cost_per_token = gpu_cost_per_hour \u002F tokens_per_hour\n    cost_per_million = cost_per_token * 1_000_000\n    return cost_per_million\n```\n\n### Reference Throughput (vLLM, single GPU)\n\n| Model Size | GPU | Tokens\u002Fsec (batch) | Cost\u002FM tokens |\n|------------|-----|-------------------|---------------|\n| 8B | A100-80GB | ~3,000-5,000 | $0.09-0.15 |\n| 8B | H100-80GB | ~5,000-8,000 | $0.11-0.18 |\n| 14B | A100-80GB | ~1,500-3,000 | $0.15-0.30 |\n| 14B | H100-80GB | ~3,000-5,000 | $0.18-0.30 |\n| 32B | H100-80GB | ~800-1,500 | $0.60-1.10 |\n| 70B | 2xH100 | ~500-1,000 | $1.80-3.60 |\n\n**Key insight**: A specialized 8B model on Modal\u002FTensorPool costs **$0.05-0.15\u002FM tokens** vs **$5-25\u002FM** for premium frontier = **50-500x cheaper per request**.\n\n## Break-Even Calculator\n\n```python\ndef break_even_analysis(\n    monthly_api_spend: float,\n    training_cost: float,\n    monthly_inference_cost: float,\n    setup_hours: float = 40,\n    hourly_rate: float = 100,\n):\n    \"\"\"Calculate break-even timeline for model specialization.\n\n    Args:\n        monthly_api_spend: Current monthly frontier API cost\n        training_cost: One-time training cost (GPU + data prep)\n        monthly_inference_cost: Monthly cost to serve specialized model\n        setup_hours: Engineering time to set up pipeline\n        hourly_rate: Engineering cost per hour\n    \"\"\"\n    engineering_cost = setup_hours * hourly_rate\n    total_upfront = training_cost + engineering_cost\n    monthly_savings = monthly_api_spend - monthly_inference_cost\n\n    if monthly_savings \u003C= 0:\n        return {\n            \"viable\": False,\n            \"reason\": \"Specialized model costs more to serve than frontier API\",\n            \"monthly_savings\": monthly_savings,\n        }\n\n    months_to_break_even = total_upfront \u002F monthly_savings\n\n    return {\n        \"viable\": True,\n        \"total_upfront_cost\": total_upfront,\n        \"training_cost\": training_cost,\n        \"engineering_cost\": engineering_cost,\n        \"monthly_api_spend\": monthly_api_spend,\n        \"monthly_inference_cost\": monthly_inference_cost,\n        \"monthly_savings\": monthly_savings,\n        \"months_to_break_even\": round(months_to_break_even, 1),\n        \"annual_savings\": monthly_savings * 12 - total_upfront,\n        \"year_2_savings\": monthly_savings * 12,  # Fully amortized\n    }\n\n\n# Example: Replace GPT-4o for a specific task\nresult = break_even_analysis(\n    monthly_api_spend=5000,       # $5K\u002Fmonth on GPT-4o\n    training_cost=500,            # $500 for LoRA fine-tune + data prep\n    monthly_inference_cost=200,   # $200\u002Fmonth on Modal (8B model)\n    setup_hours=40,               # 1 week of engineering\n    hourly_rate=100,              # $100\u002Fhr engineer\n)\n# Result: Break-even in ~1.1 months, $52K annual savings after year 1\n```\n\n## ROI Timeline Template\n\n| Month | Frontier Cost | Specialized Cost | Cumulative Savings | Notes |\n|-------|--------------|-----------------|-------------------|-------|\n| 0 | — | $4,500 upfront | -$4,500 | Training + engineering |\n| 1 | $5,000 | $200 | $300 | First month live |\n| 2 | $5,000 | $200 | $5,100 | Break-even |\n| 3 | $5,000 | $200 | $9,900 | |\n| 6 | $5,000 | $200 | $24,300 | |\n| 12 | $5,000 | $300 | $52,200 | Includes retrain cost |\n\n## Decision Framework\n\n### Strong Case for Specialization\n- Monthly API spend > $1,000\n- Constrained task (classification, extraction, formatting, domain Q&A)\n- Production data available (logs, corrections, accept\u002Freject signals)\n- Quality requirements are well-defined and measurable\n- Task doesn't change frequently\n\n### Weak Case for Specialization\n- Monthly API spend \u003C $500\n- Diverse, open-ended tasks (general assistant)\n- No production data to train on\n- Rapid model improvement expected (new frontier models releasing soon)\n- Task requires reasoning at the frontier level\n- Small team with no ML engineering capacity\n\n### Hybrid Approach\nRoute easy requests to specialized model, hard ones to frontier:\n- **90% of requests** → Specialized 8B (cheap, fast)\n- **10% of requests** → Frontier fallback (complex, novel)\n- **Result**: 80-90% cost reduction while maintaining quality on hard cases\n\n## Quick Assessment Questions\n\nAsk the user these to determine viability:\n\n1. **What's your monthly frontier API spend?**\n   - \u003C $500: Probably not worth it (low ROI)\n   - $500-2K: Worth investigating\n   - $2K+: Strong candidate\n   - $10K+: Almost certainly worth it\n\n2. **How constrained is the task?**\n   - Single task (classification, extraction): Great fit\n   - Few related tasks: Good fit\n   - Many diverse tasks: Harder to specialize\n\n3. **What production data do you have?**\n   - API logs with user feedback: Excellent\n   - API logs without feedback: Good (can distill)\n   - No production data: Need synthetic bootstrap\n\n4. **What's your quality bar?**\n   - Must match frontier exactly: Harder, may need bigger model\n   - 90% of frontier quality is fine: 8B LoRA likely sufficient\n   - Task-specific metrics (accuracy, format): Easiest to optimize\n",{"name":5,"tags":44,"author":52,"license":15,"version":53,"category":54,"description":7,"dependencies":55},[45,46,47,48,23,49,50,51],"Economics","Cost Analysis","ROI","Break-Even","Inference Cost","Model Flywheel","Business Case","Synthetic Sciences","1.0.0","ml-training",[],"main","4aa39efd1cdab0b4657ecee16c92a92011a5b10f","openscience\u002Fbackend\u002Fcli\u002Fskills\u002Fml-training\u002Fmodel-economics","https:\u002F\u002Fgithub.com\u002Ffadrienne\u002FMyGitHub","https:\u002F\u002Fgithub.com\u002Ffadrienne","2026-10-28 00:43:09",{"source":8,"skillName":5,"commands":63,"prompt":73,"zipAvailable":74},[64,67,70],{"tool":65,"command":66},"npx","npx skills add fadrienne\u002FMyGitHub --skill model-economics",{"tool":68,"command":69},"bunx","bunx skills add fadrienne\u002FMyGitHub --skill model-economics",{"tool":71,"command":72},"pnpm","pnpm dlx skills add fadrienne\u002FMyGitHub --skill model-economics","安装技能，技能地址：https:\u002F\u002Fgithub.com\u002Ffadrienne\u002FMyGitHub\u002Fblob\u002Fmain\u002Fopenscience\u002Fbackend\u002Fcli\u002Fskills\u002Fml-training\u002Fmodel-economics\u002FSKILL.md",true,[],[77,107,133,167,194,221],{"skillNo":78,"name":79,"nameZh":6,"description":80,"descriptionZh":6,"repositoryFullName":8,"ownerLogin":9,"ownerAvatarUrl":10,"ownerAccountNo":11,"ownerDisplayName":12,"repositoryNo":13,"starsCount":14,"licenseSpdxId":15,"licenseFileUrl":6,"sourceHtmlUrl":81,"categories":82,"keywords":88,"publishedAt":39},"01M7167B28PK20HP7R7ZED2DY0","statistical-power","Sample size calculation, power analysis, and effect size conversion for scientific studies. Use BEFORE data collection (a priori) to determine required N, or after (post-hoc) to compute achieved power. Covers parametric tests (t-test, ANOVA, chi-square, proportions, correlation, regression), simulation-based power for complex\u002Fclustered\u002Fnested designs, and power curves. Use this skill when someone asks 'how many samples do I need', 'what is my power', 'what effect size can I detect', or any variant. Pairs with \u002Fexperimental-design (choose design first, then compute power) and \u002Fstatistical-analysis (analyze data after collection).","https:\u002F\u002Fgithub.com\u002Ffadrienne\u002FMyGitHub\u002Fblob\u002Fmain\u002Fscientific-agent-skills\u002Fskills\u002Fstatistical-power\u002FSKILL.md",[83],{"categoryNo":84,"categoryName":85,"taxonomyType":86,"taxonomyTypeDesc":87},"01JBSEEDCAT000000000000023","Skill与智能体开发","SCENARIO","使用场景",[89,90,91,92,93,94,95,96,24,97,98,35,99,100,101,102,103,104,105,106],"power","statistical power","effect","scipy","regression","choose","design","uv","Bash","collection","calculation","ANOVA","conversion","analyze","t-test","chi-square","simulation","sample",{"skillNo":108,"name":109,"nameZh":6,"description":110,"descriptionZh":6,"repositoryFullName":8,"ownerLogin":9,"ownerAvatarUrl":10,"ownerAccountNo":11,"ownerDisplayName":12,"repositoryNo":13,"starsCount":14,"licenseSpdxId":15,"licenseFileUrl":6,"sourceHtmlUrl":111,"categories":112,"keywords":116,"publishedAt":39},"01M71677EBCM8EPCE9BZCK8505","statistical-analysis","Guided statistical analysis with test selection and reporting. Use when you need help choosing appropriate tests for your data, assumption checking, power analysis, and APA-formatted results. Best for academic research reporting, test selection guidance. For implementing specific models programmatically use statsmodels.","https:\u002F\u002Fgithub.com\u002Ffadrienne\u002FMyGitHub\u002Fblob\u002Fmain\u002Fscientific-agent-skills\u002Fskills\u002Fstatistical-analysis\u002FSKILL.md",[113],{"categoryNo":114,"categoryName":115,"taxonomyType":86,"taxonomyTypeDesc":87},"01JBSEEDCAT000000000000008","测试与质量",[35,117,89,118,119,120,24,121,122,123,27,124,125,126,127,128,129,130,131,132],"statistical analysis","academic research","implementing","selection","you","reporting","data","tests","use","with","test","research","results","statsmodels","models","help",{"skillNo":134,"name":135,"nameZh":6,"description":136,"descriptionZh":6,"repositoryFullName":8,"ownerLogin":9,"ownerAvatarUrl":10,"ownerAccountNo":11,"ownerDisplayName":12,"repositoryNo":13,"starsCount":14,"licenseSpdxId":15,"licenseFileUrl":6,"sourceHtmlUrl":137,"categories":138,"keywords":148,"publishedAt":39},"01M71673R93XAHFH81DBNVNG7H","scientific-writing","Core skill for the deep research and writing tool. Write scientific manuscripts in full paragraphs (never bullet points). Use two-stage process with (1) section outlines with key points using research-lookup then (2) convert to flowing prose. IMRAD structure, citations (APA\u002FAMA\u002FVancouver), figures\u002Ftables, reporting guidelines (CONSORT\u002FSTROBE\u002FPRISMA), for research papers and journal submissions.","https:\u002F\u002Fgithub.com\u002Ffadrienne\u002FMyGitHub\u002Fblob\u002Fmain\u002Fscientific-agent-skills\u002Fskills\u002Fscientific-writing\u002FSKILL.md",[139,142,143],{"categoryNo":140,"categoryName":141,"taxonomyType":86,"taxonomyTypeDesc":87},"01JBSEEDCAT000000000000018","写作与文案",{"categoryNo":84,"categoryName":85,"taxonomyType":86,"taxonomyTypeDesc":87},{"categoryNo":144,"categoryName":145,"taxonomyType":146,"taxonomyTypeDesc":147},"01JBSEEDCAT000000000000036","内容创作者","PROFESSION","适用职业",[149,150,151,152,153,154,155,156,157,158,97,159,160,161,162,163,164,165,122,166],"writing","scientific","scientific writing","Citations","deep","deep research","never","papers","journal","false","api","prisma","structure","openrouter","Outlines","core skill","read","features",{"skillNo":168,"name":169,"nameZh":6,"description":170,"descriptionZh":6,"repositoryFullName":8,"ownerLogin":9,"ownerAvatarUrl":10,"ownerAccountNo":11,"ownerDisplayName":12,"repositoryNo":13,"starsCount":14,"licenseSpdxId":15,"licenseFileUrl":6,"sourceHtmlUrl":171,"categories":172,"keywords":177,"publishedAt":39},"01M7166ZPKXBF4EBPHVTJPQBB1","scientific-visualization","Meta-skill for publication-ready figures. Use when creating journal submission figures requiring multi-panel layouts, significance annotations, error bars, colorblind-safe palettes, and specific journal formatting (Nature, Science, Cell). Orchestrates matplotlib\u002Fseaborn\u002Fplotly with publication styles. For quick exploration use seaborn or plotly directly.","https:\u002F\u002Fgithub.com\u002Ffadrienne\u002FMyGitHub\u002Fblob\u002Fmain\u002Fscientific-agent-skills\u002Fskills\u002Fscientific-visualization\u002FSKILL.md",[173,176],{"categoryNo":174,"categoryName":175,"taxonomyType":86,"taxonomyTypeDesc":87},"01JBSEEDCAT000000000000013","分析与可视化",{"categoryNo":84,"categoryName":85,"taxonomyType":86,"taxonomyTypeDesc":87},[178,150,179,180,181,182,183,184,185,157,24,186,187,188,189,190,191,192,27,193],"visualization","scientific visualization","error","ready","quick","exploration","meta-skill","Meta","safe","plotly","annotations","formatting","Publication","figures","panel","submission",{"skillNo":195,"name":196,"nameZh":6,"description":197,"descriptionZh":6,"repositoryFullName":8,"ownerLogin":9,"ownerAvatarUrl":10,"ownerAccountNo":11,"ownerDisplayName":12,"repositoryNo":13,"starsCount":14,"licenseSpdxId":15,"licenseFileUrl":6,"sourceHtmlUrl":198,"categories":199,"keywords":206,"publishedAt":39},"01M7166VXJWQSSC3RQZP60WBGR","research-code","Debug, profile, review, and restructure Python research and analysis code. Use when analysis scripts crash, produce wrong results, run too slowly, or need a correctness review. Covers: reading tracebacks and diagnosing root causes, hypothesis-driven debugging, reviewing code for common statistical\u002Fnumerical errors, profiling data pipelines, improving reproducibility (seeds, environments, determinism), and structuring analysis scripts for sharing. Distinct from \u002Fexploratory-data-analysis (which understands data) and \u002Fstatistical-analysis (which chooses and runs tests) — this skill focuses on the code itself.","https:\u002F\u002Fgithub.com\u002Ffadrienne\u002FMyGitHub\u002Fblob\u002Fmain\u002Fscientific-agent-skills\u002Fskills\u002Fresearch-code\u002FSKILL.md",[200,203],{"categoryNo":201,"categoryName":202,"taxonomyType":86,"taxonomyTypeDesc":87},"01JBSEEDCAT000000000000021","资料检索与研究",{"categoryNo":204,"categoryName":205,"taxonomyType":146,"taxonomyTypeDesc":147},"01JBSEEDCAT000000000000030","开发工程师",[207,208,128,209,210,211,212,96,213,24,97,35,214,215,216,217,218,219,165,220],"code","research code","which","memory","data-analysis","errors","debugging","common","review","data pipelines","run","scripts","line","write",{"skillNo":222,"name":223,"nameZh":6,"description":224,"descriptionZh":6,"repositoryFullName":8,"ownerLogin":9,"ownerAvatarUrl":10,"ownerAccountNo":11,"ownerDisplayName":12,"repositoryNo":13,"starsCount":14,"licenseSpdxId":225,"licenseFileUrl":6,"sourceHtmlUrl":226,"categories":227,"keywords":234,"publishedAt":39},"01M7166CF5HSR7SVZYZFWTDX55","markdown-mermaid-writing","Comprehensive markdown and Mermaid diagram writing skill. Use when creating any scientific document, report, analysis, or visualization. Establishes text-based diagrams as the default documentation standard with full style guides (markdown + mermaid), 24 diagram type references, and 9 document templates.","Apache-2.0","https:\u002F\u002Fgithub.com\u002Ffadrienne\u002FMyGitHub\u002Fblob\u002Fmain\u002Fscientific-agent-skills\u002Fskills\u002Fmarkdown-mermaid-writing\u002FSKILL.md",[228,231,232,233],{"categoryNo":229,"categoryName":230,"taxonomyType":86,"taxonomyTypeDesc":87},"01JBSEEDCAT000000000000002","文档处理",{"categoryNo":140,"categoryName":141,"taxonomyType":86,"taxonomyTypeDesc":87},{"categoryNo":84,"categoryName":85,"taxonomyType":86,"taxonomyTypeDesc":87},{"categoryNo":144,"categoryName":145,"taxonomyType":146,"taxonomyTypeDesc":147},[149,235,236,237,238,239,178,240,241,24,97,35,242,243,244,245,246,247,150,165],"markdown mermaid writing","mermaid","markdown","community","diagrams","project","default","team","guides","text","documentation","references","diagram"]