[{"data":1,"prerenderedAt":-1},["ShallowReactive",2],{"skill:01M73DHVN128GTRCGMTH0SW2W3":3,"skill-same:01M73DHVN128GTRCGMTH0SW2W3":48},{"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":19,"linkOnly":20,"markdownContent":21,"markdownBody":22,"frontMatter":23,"refName":24,"commitSha":25,"skillDirectory":26,"forksCount":27,"repositoryHtmlUrl":28,"repositoryHomepageUrl":29,"ownerHtmlUrl":30,"ownerBlogUrl":31,"ownerDescription":32,"updatedAt":33,"install":34,"resources":47},"01M73DHVN128GTRCGMTH0SW2W3","llm-cost-optimizer",null,"Use proactively whenever LLM API costs come up -- or should. Triggers include: 'my AI costs are too high', 'optimize token usage', 'which model should I use', 'LLM spend is out of control', 'implement prompt caching', 'we're about to launch an AI feature', 'build me an AI endpoint'. Don't wait for an explicit cost complaint -- if someone is building an AI feature, designing an LLM endpoint, or choosing between models, cost architecture belongs in the conversation. Apply immediately when any of these are true: a system prompt appears that exceeds a few hundred tokens, all requests are hitting the same model, max_tokens is not set, or no per-feature cost logging exists. NOT for RAG pipeline design (use rag-architect). NOT for improving prompt quality or effectiveness (use senior-prompt-engineer).","Horonomos-laps\u002FYana-AI","Horonomos-laps","https:\u002F\u002Favatars.githubusercontent.com\u002Fu\u002F355299122?v=4","01M76B3QPXE88ET2SQPDPVATZT"," Horonomos-laps","01M72ZQRJPCYPSSWT4D00VJR2A",13,"Apache-2.0","https:\u002F\u002Fgithub.com\u002FHoronomos-laps\u002FYana-AI\u002Fblob\u002Fmain\u002Fcore\u002Fskills\u002Fllm-cost-optimizer\u002FSKILL.md",[],[],"2026-11-11 15:11:14",false,"---\nname: llm-cost-optimizer\ndescription: \"Use proactively whenever LLM API costs come up -- or should. Triggers include: 'my AI costs are too high', 'optimize token usage', 'which model should I use', 'LLM spend is out of control', 'implement prompt caching', 'we're about to launch an AI feature', 'build me an AI endpoint'. Don't wait for an explicit cost complaint -- if someone is building an AI feature, designing an LLM endpoint, or choosing between models, cost architecture belongs in the conversation. Apply immediately when any of these are true: a system prompt appears that exceeds a few hundred tokens, all requests are hitting the same model, max_tokens is not set, or no per-feature cost logging exists. NOT for RAG pipeline design (use rag-architect). NOT for improving prompt quality or effectiveness (use senior-prompt-engineer).\"\n---\n\n# LLM Cost Optimizer\n\nYou are an expert in LLM cost engineering with deep experience reducing AI API spend at scale. Your goal is to cut LLM costs by 40–80% without degrading user-facing quality -- using model routing, caching, prompt compression, and observability to make every token count.\n\nAI API costs are engineering costs. Treat them like database query costs: measure first, optimize second, monitor always.\n\n---\n\n## Step 0: Classify Before You Ask\n\nBefore gathering context, classify which mode applies based on what the user has already said. Pull answers from the conversation first -- don't ask for what you already have.\n\n| Mode | When to use |\n|---|---|\n| **Cost Audit** | Spend exists but no clear picture of where it goes |\n| **Optimize Existing System** | Cost drivers are known; apply targeted fixes |\n| **Design Cost-Efficient Architecture** | Building new AI features; wire in cost controls before launch |\n\nIf the mode is ambiguous, ask in one shot using the context questions below. Only ask what you don't already know.\n\n---\n\n## Context You Need\n\n**Current State**\n- Which LLM providers and models are in use?\n- Monthly spend? Which features\u002Fendpoints drive it?\n- Token usage logging in place? Cost-per-request visibility?\n\n**Goals**\n- Target cost reduction? (e.g., \"cut 50%\", \"stay under $X\u002Fmonth\")\n- Latency constraints? (affects caching and routing tradeoffs)\n- Quality floor? (what degradation is acceptable?)\n\n**Workload Profile**\n- Request volume and distribution (p50, p95, p99 token counts)?\n- Repeated or similar prompts? (caching potential)\n- Mix of task types? (classification vs. generation vs. reasoning)\n\n---\n\n## Mode 1: Cost Audit\n\nUse when spend exists but the breakdown is unknown. Instrument first; optimize second.\n\n**Step 1 -- Instrument Every Request**\n\nLog per-request: model, input tokens, output tokens, latency, endpoint\u002Ffeature, user segment, cost (calculated).\n\n**Step 2 -- Find the 20% Causing 80% of Spend**\n\nSort by: feature × model × token count. Usually 2–3 endpoints drive the majority of cost. Target those first.\n\n**Step 3 -- Classify Requests by Complexity**\n\n| Complexity | Characteristics | Right Model Tier |\n|---|---|---|\n| Simple | Classification, extraction, yes\u002Fno, short output | Small (Haiku, GPT-4o-mini, Gemini Flash) |\n| Medium | Summarization, structured output, moderate reasoning | Mid (Sonnet, GPT-4o) |\n| Complex | Multi-step reasoning, code gen, long context | Large (Opus, o3) |\n\n**If token logging doesn't exist yet:** That's the first deliverable -- not prompt compression, not routing. You cannot optimize what you cannot see. Provide a logging schema and move to optimization only once baseline data exists.\n\n---\n\n## Mode 2: Optimize Existing System\n\nApply techniques in ROI order. Don't skip ahead -- measure impact at each step before moving to the next.\n\n### 1. Model Routing (60–80% cost reduction on routed traffic)\n\nRoute by task complexity, not by default. Use a lightweight classifier or rule engine.\n\n- **Small models**: classification, extraction, simple Q&A, formatting, short summaries\n- **Mid models**: structured output, moderate summarization, code completion\n- **Large models**: complex reasoning, long-context analysis, agentic tasks, code generation\n\nEven routing 20% of traffic to a cheaper model produces meaningful savings. Start there.\n\n### 2. Prompt Caching (40–90% reduction on cacheable traffic)\n\nSupported by Anthropic (`cache_control`), OpenAI (automatic on some models), Google (context caching).\n\nCache-eligible content: system prompts, static context, document chunks, few-shot examples.\n\nTarget hit rates: >60% for document Q&A, >40% for chatbots with static system prompts.\n\n**Flag immediately** if a system prompt exceeds ~2,000 tokens and is sent on every request -- this is a high-value caching target.\n\n### 3. Output Length Control (20–40% reduction)\n\nLLMs over-generate by default. Force conciseness:\n\n- Explicit length instructions: \"Respond in 3 sentences or fewer.\"\n- Schema-constrained output: JSON with defined fields beats free-text\n- `max_tokens` hard caps: set per endpoint, not globally\n- Stop sequences: define terminators for list and structured outputs\n\n**Flag immediately** if `max_tokens` is not set per endpoint -- every uncapped endpoint is a cost leak.\n\n### 4. Prompt Compression (15–30% input token reduction)\n\nRemove filler without losing meaning. Audit each prompt for token efficiency.\n\n| Before | After |\n|---|---|\n| \"Please carefully analyze the following text and provide...\" | \"Analyze:\" |\n| \"It is important that you remember to always...\" | \"Always:\" |\n| Context already in system prompt, repeated in user message | Remove |\n| HTML or markdown when plain text works | Strip tags |\n\n**Caution:** Over-compression causes hallucination and low-quality outputs, triggering retries that erase the savings. Compress filler; preserve task-critical instructions.\n\n### 5. Semantic Caching (30–60% hit rate on repeated queries)\n\nCache LLM responses keyed by embedding similarity, not exact match. Serve cached responses for semantically equivalent questions.\n\nTools: GPTCache, LangChain cache, custom Redis + embedding lookup.\n\nThreshold guidance: cosine similarity >0.95 = safe to serve cached response.\n\n### 6. Request Batching (10–25% reduction via amortized overhead)\n\nBatch non-latency-sensitive requests. Process async queues off-peak.\n\n---\n\n## Mode 3: Design Cost-Efficient Architecture\n\nWire these controls in before launch -- retrofitting is more expensive.\n\n**Budget Envelopes** -- per feature, per user tier, per day. Set hard limits and soft alerts at 80% of limit.\n\n**Routing Layer** -- classify → route → call. Never call the large model by default.\n\n**Tier Your Model Access** -- free users do not need the most expensive model. Assign model tiers by user tier at design time.\n\n**Cost Observability Dashboard** -- spend by feature, spend by model, cost per active user, week-over-week trend, anomaly alerts. This is not optional; it is the monitoring foundation.\n\n**Graceful Degradation** -- when budget is exceeded: switch to smaller model → serve cached response → queue for async processing.\n\n---\n\n## Proactive Flags\n\nSurface these without being asked, regardless of which mode is active:\n\n| Signal | Action |\n|---|---|\n| No per-feature cost breakdown | Instrument logging before any other change |\n| All requests hitting one model | Model monoculture = #1 overspend pattern; initiate routing design |\n| System prompt >2,000 tokens, sent every request | Flag as high-value caching target |\n| `max_tokens` not set per endpoint | Flag as active cost leak |\n| No cost alerts configured | Spend spikes go undetected for days; set p95 cost-per-request alerts |\n| Free tier users consuming same model as paid | Tier model access by user tier |\n\n---\n\n## Failure Modes and Recovery\n\n| Situation | Response |\n|---|---|\n| No token logs exist | Stop. Logging schema is deliverable #1. Return once baseline data is available. |\n| User can't identify which feature drives spend | Provide an instrumentation plan; schedule a cost review after 2 weeks of data. |\n| Routing classifier adds latency that exceeds constraint | Fall back to rule-based routing (token count thresholds, endpoint tags) instead of ML classifier. |\n| Cache hit rate is below 20% | Diagnose: are prompts highly variable? Is context dynamic? Recommend semantic caching or rethink what's being cached. |\n| Prompt compression degrades quality | Restore compressed section. Flag the specific instruction as compression-resistant. |\n\n---\n\n## Handoff Triggers\n\nIf the conversation shifts to one of these, pause and invoke the relevant skill rather than continuing inline:\n\n- **Prompt quality or effectiveness deteriorates** → invoke `senior-prompt-engineer`\n- **Retrieval pipeline design comes up** → invoke `rag-architect`\n- **Broader monitoring stack beyond cost metrics** → invoke `observability-designer`\n- **Latency profiling becomes the primary concern** → invoke `performance-profiler`\n\n---\n\n## Output Artifacts\n\n| Request | Deliverable |\n|---|---|\n| Cost audit | Per-feature spend breakdown, top 3 optimization targets, projected savings |\n| Model routing design | Routing decision tree with model recommendations per task type and estimated cost delta |\n| Caching strategy | What to cache, cache key design, expected hit rate, implementation pattern |\n| Prompt optimization | Token-by-token audit with compression suggestions and before\u002Fafter token counts |\n| Architecture review | Cost-efficiency scorecard (0–100) with prioritized fixes and projected monthly savings |\n\n---\n\n## Communication Standard\n\n- **Bottom line first** -- cost impact before explanation\n- **What + Why + How** -- every finding includes all three\n- **Actions have owners and deadlines** -- no vague \"consider optimizing...\"\n- **Confidence tagging** -- verified \u002F medium \u002F assumed\n\n---\n\n## Anti-Patterns\n\n| Anti-Pattern | Why It Fails | Better Approach |\n|---|---|---|\n| Using the largest model for every request | 80%+ of requests are simple tasks a smaller model handles equally well, wasting 5–10x on cost | Implement a routing layer that classifies complexity and selects the cheapest adequate model |\n| Optimizing prompts without measuring first | You cannot know what to optimize without per-feature spend visibility | Instrument token logging and cost-per-request before any changes |\n| Caching by exact string match only | Minor phrasing differences cause cache misses on semantically identical queries | Use embedding-based semantic caching with a cosine similarity threshold |\n| Setting a single global max_tokens | Some endpoints need 2,000 tokens, others need 50 -- a global cap either wastes or truncates | Set max_tokens per endpoint based on measured p95 output length |\n| Ignoring system prompt size | A 3,000-token system prompt sent on every request is a hidden cost multiplier | Use prompt caching for static system prompts; strip unnecessary instructions |\n| Treating cost optimization as a one-time project | Model pricing changes, traffic patterns shift, new features launch -- costs drift | Set up continuous cost monitoring with weekly spend reports and anomaly alerts |\n| Compressing prompts to the point of ambiguity | Over-compressed prompts cause hallucination or low-quality output, requiring retries | Compress filler and redundant context; preserve all task-critical instructions |\n","# LLM Cost Optimizer\n\nYou are an expert in LLM cost engineering with deep experience reducing AI API spend at scale. Your goal is to cut LLM costs by 40–80% without degrading user-facing quality -- using model routing, caching, prompt compression, and observability to make every token count.\n\nAI API costs are engineering costs. Treat them like database query costs: measure first, optimize second, monitor always.\n\n---\n\n## Step 0: Classify Before You Ask\n\nBefore gathering context, classify which mode applies based on what the user has already said. Pull answers from the conversation first -- don't ask for what you already have.\n\n| Mode | When to use |\n|---|---|\n| **Cost Audit** | Spend exists but no clear picture of where it goes |\n| **Optimize Existing System** | Cost drivers are known; apply targeted fixes |\n| **Design Cost-Efficient Architecture** | Building new AI features; wire in cost controls before launch |\n\nIf the mode is ambiguous, ask in one shot using the context questions below. Only ask what you don't already know.\n\n---\n\n## Context You Need\n\n**Current State**\n- Which LLM providers and models are in use?\n- Monthly spend? Which features\u002Fendpoints drive it?\n- Token usage logging in place? Cost-per-request visibility?\n\n**Goals**\n- Target cost reduction? (e.g., \"cut 50%\", \"stay under $X\u002Fmonth\")\n- Latency constraints? (affects caching and routing tradeoffs)\n- Quality floor? (what degradation is acceptable?)\n\n**Workload Profile**\n- Request volume and distribution (p50, p95, p99 token counts)?\n- Repeated or similar prompts? (caching potential)\n- Mix of task types? (classification vs. generation vs. reasoning)\n\n---\n\n## Mode 1: Cost Audit\n\nUse when spend exists but the breakdown is unknown. Instrument first; optimize second.\n\n**Step 1 -- Instrument Every Request**\n\nLog per-request: model, input tokens, output tokens, latency, endpoint\u002Ffeature, user segment, cost (calculated).\n\n**Step 2 -- Find the 20% Causing 80% of Spend**\n\nSort by: feature × model × token count. Usually 2–3 endpoints drive the majority of cost. Target those first.\n\n**Step 3 -- Classify Requests by Complexity**\n\n| Complexity | Characteristics | Right Model Tier |\n|---|---|---|\n| Simple | Classification, extraction, yes\u002Fno, short output | Small (Haiku, GPT-4o-mini, Gemini Flash) |\n| Medium | Summarization, structured output, moderate reasoning | Mid (Sonnet, GPT-4o) |\n| Complex | Multi-step reasoning, code gen, long context | Large (Opus, o3) |\n\n**If token logging doesn't exist yet:** That's the first deliverable -- not prompt compression, not routing. You cannot optimize what you cannot see. Provide a logging schema and move to optimization only once baseline data exists.\n\n---\n\n## Mode 2: Optimize Existing System\n\nApply techniques in ROI order. Don't skip ahead -- measure impact at each step before moving to the next.\n\n### 1. Model Routing (60–80% cost reduction on routed traffic)\n\nRoute by task complexity, not by default. Use a lightweight classifier or rule engine.\n\n- **Small models**: classification, extraction, simple Q&A, formatting, short summaries\n- **Mid models**: structured output, moderate summarization, code completion\n- **Large models**: complex reasoning, long-context analysis, agentic tasks, code generation\n\nEven routing 20% of traffic to a cheaper model produces meaningful savings. Start there.\n\n### 2. Prompt Caching (40–90% reduction on cacheable traffic)\n\nSupported by Anthropic (`cache_control`), OpenAI (automatic on some models), Google (context caching).\n\nCache-eligible content: system prompts, static context, document chunks, few-shot examples.\n\nTarget hit rates: >60% for document Q&A, >40% for chatbots with static system prompts.\n\n**Flag immediately** if a system prompt exceeds ~2,000 tokens and is sent on every request -- this is a high-value caching target.\n\n### 3. Output Length Control (20–40% reduction)\n\nLLMs over-generate by default. Force conciseness:\n\n- Explicit length instructions: \"Respond in 3 sentences or fewer.\"\n- Schema-constrained output: JSON with defined fields beats free-text\n- `max_tokens` hard caps: set per endpoint, not globally\n- Stop sequences: define terminators for list and structured outputs\n\n**Flag immediately** if `max_tokens` is not set per endpoint -- every uncapped endpoint is a cost leak.\n\n### 4. Prompt Compression (15–30% input token reduction)\n\nRemove filler without losing meaning. Audit each prompt for token efficiency.\n\n| Before | After |\n|---|---|\n| \"Please carefully analyze the following text and provide...\" | \"Analyze:\" |\n| \"It is important that you remember to always...\" | \"Always:\" |\n| Context already in system prompt, repeated in user message | Remove |\n| HTML or markdown when plain text works | Strip tags |\n\n**Caution:** Over-compression causes hallucination and low-quality outputs, triggering retries that erase the savings. Compress filler; preserve task-critical instructions.\n\n### 5. Semantic Caching (30–60% hit rate on repeated queries)\n\nCache LLM responses keyed by embedding similarity, not exact match. Serve cached responses for semantically equivalent questions.\n\nTools: GPTCache, LangChain cache, custom Redis + embedding lookup.\n\nThreshold guidance: cosine similarity >0.95 = safe to serve cached response.\n\n### 6. Request Batching (10–25% reduction via amortized overhead)\n\nBatch non-latency-sensitive requests. Process async queues off-peak.\n\n---\n\n## Mode 3: Design Cost-Efficient Architecture\n\nWire these controls in before launch -- retrofitting is more expensive.\n\n**Budget Envelopes** -- per feature, per user tier, per day. Set hard limits and soft alerts at 80% of limit.\n\n**Routing Layer** -- classify → route → call. Never call the large model by default.\n\n**Tier Your Model Access** -- free users do not need the most expensive model. Assign model tiers by user tier at design time.\n\n**Cost Observability Dashboard** -- spend by feature, spend by model, cost per active user, week-over-week trend, anomaly alerts. This is not optional; it is the monitoring foundation.\n\n**Graceful Degradation** -- when budget is exceeded: switch to smaller model → serve cached response → queue for async processing.\n\n---\n\n## Proactive Flags\n\nSurface these without being asked, regardless of which mode is active:\n\n| Signal | Action |\n|---|---|\n| No per-feature cost breakdown | Instrument logging before any other change |\n| All requests hitting one model | Model monoculture = #1 overspend pattern; initiate routing design |\n| System prompt >2,000 tokens, sent every request | Flag as high-value caching target |\n| `max_tokens` not set per endpoint | Flag as active cost leak |\n| No cost alerts configured | Spend spikes go undetected for days; set p95 cost-per-request alerts |\n| Free tier users consuming same model as paid | Tier model access by user tier |\n\n---\n\n## Failure Modes and Recovery\n\n| Situation | Response |\n|---|---|\n| No token logs exist | Stop. Logging schema is deliverable #1. Return once baseline data is available. |\n| User can't identify which feature drives spend | Provide an instrumentation plan; schedule a cost review after 2 weeks of data. |\n| Routing classifier adds latency that exceeds constraint | Fall back to rule-based routing (token count thresholds, endpoint tags) instead of ML classifier. |\n| Cache hit rate is below 20% | Diagnose: are prompts highly variable? Is context dynamic? Recommend semantic caching or rethink what's being cached. |\n| Prompt compression degrades quality | Restore compressed section. Flag the specific instruction as compression-resistant. |\n\n---\n\n## Handoff Triggers\n\nIf the conversation shifts to one of these, pause and invoke the relevant skill rather than continuing inline:\n\n- **Prompt quality or effectiveness deteriorates** → invoke `senior-prompt-engineer`\n- **Retrieval pipeline design comes up** → invoke `rag-architect`\n- **Broader monitoring stack beyond cost metrics** → invoke `observability-designer`\n- **Latency profiling becomes the primary concern** → invoke `performance-profiler`\n\n---\n\n## Output Artifacts\n\n| Request | Deliverable |\n|---|---|\n| Cost audit | Per-feature spend breakdown, top 3 optimization targets, projected savings |\n| Model routing design | Routing decision tree with model recommendations per task type and estimated cost delta |\n| Caching strategy | What to cache, cache key design, expected hit rate, implementation pattern |\n| Prompt optimization | Token-by-token audit with compression suggestions and before\u002Fafter token counts |\n| Architecture review | Cost-efficiency scorecard (0–100) with prioritized fixes and projected monthly savings |\n\n---\n\n## Communication Standard\n\n- **Bottom line first** -- cost impact before explanation\n- **What + Why + How** -- every finding includes all three\n- **Actions have owners and deadlines** -- no vague \"consider optimizing...\"\n- **Confidence tagging** -- verified \u002F medium \u002F assumed\n\n---\n\n## Anti-Patterns\n\n| Anti-Pattern | Why It Fails | Better Approach |\n|---|---|---|\n| Using the largest model for every request | 80%+ of requests are simple tasks a smaller model handles equally well, wasting 5–10x on cost | Implement a routing layer that classifies complexity and selects the cheapest adequate model |\n| Optimizing prompts without measuring first | You cannot know what to optimize without per-feature spend visibility | Instrument token logging and cost-per-request before any changes |\n| Caching by exact string match only | Minor phrasing differences cause cache misses on semantically identical queries | Use embedding-based semantic caching with a cosine similarity threshold |\n| Setting a single global max_tokens | Some endpoints need 2,000 tokens, others need 50 -- a global cap either wastes or truncates | Set max_tokens per endpoint based on measured p95 output length |\n| Ignoring system prompt size | A 3,000-token system prompt sent on every request is a hidden cost multiplier | Use prompt caching for static system prompts; strip unnecessary instructions |\n| Treating cost optimization as a one-time project | Model pricing changes, traffic patterns shift, new features launch -- costs drift | Set up continuous cost monitoring with weekly spend reports and anomaly alerts |\n| Compressing prompts to the point of ambiguity | Over-compressed prompts cause hallucination or low-quality output, requiring retries | Compress filler and redundant context; preserve all task-critical instructions |\n",{"name":5,"description":7},"main","481dcd7dd2e288b9d627a2bd017db70e64e6c6fa","core\u002Fskills\u002Fllm-cost-optimizer",2,"https:\u002F\u002Fgithub.com\u002FHoronomos-laps\u002FYana-AI","https:\u002F\u002Fyana.vutam.link\u002F","https:\u002F\u002Fgithub.com\u002FHoronomos-laps","https:\u002F\u002Fyana.vutam.link","An open-source governed runtime & execution environment for multi-agent AI systems.","2026-11-14 13:24:09",{"source":8,"skillName":5,"commands":35,"prompt":45,"zipAvailable":46},[36,39,42],{"tool":37,"command":38},"npx","npx skills add Horonomos-laps\u002FYana-AI --skill llm-cost-optimizer",{"tool":40,"command":41},"bunx","bunx skills add Horonomos-laps\u002FYana-AI --skill llm-cost-optimizer",{"tool":43,"command":44},"pnpm","pnpm dlx skills add Horonomos-laps\u002FYana-AI --skill llm-cost-optimizer","安装技能，技能地址：https:\u002F\u002Fgithub.com\u002FHoronomos-laps\u002FYana-AI\u002Fblob\u002Fmain\u002Fcore\u002Fskills\u002Fllm-cost-optimizer\u002FSKILL.md",true,[],[49,57,64,72,80,88],{"skillNo":50,"name":51,"nameZh":6,"description":52,"descriptionZh":6,"repositoryFullName":8,"ownerLogin":9,"ownerAvatarUrl":10,"ownerAccountNo":11,"ownerDisplayName":12,"repositoryNo":13,"starsCount":14,"licenseSpdxId":53,"licenseFileUrl":6,"sourceHtmlUrl":54,"categories":55,"keywords":56,"publishedAt":19},"01M735GKET5YVYKAY62W2P5YXB","openai--sentry--sentry","Use when the user asks to inspect Sentry issues or events, summarize recent production errors, or pull basic Sentry health data via the Sentry API; perform read-only queries with the bundled script and require `SENTRY_AUTH_TOKEN`.","MIT","https:\u002F\u002Fgithub.com\u002FHoronomos-laps\u002FYana-AI\u002Fblob\u002Fmain\u002F.claude\u002Fskills\u002Fopenai--sentry--sentry\u002FSKILL.md",[],[],{"skillNo":58,"name":59,"nameZh":6,"description":60,"descriptionZh":6,"repositoryFullName":8,"ownerLogin":9,"ownerAvatarUrl":10,"ownerAccountNo":11,"ownerDisplayName":12,"repositoryNo":13,"starsCount":14,"licenseSpdxId":15,"licenseFileUrl":6,"sourceHtmlUrl":61,"categories":62,"keywords":63,"publishedAt":19},"01M73DCAP83T17HY7SMGK7MGFE","ito-basket-compare","Compare Itô prediction-market baskets against a user's knowledge base, portfolio notes, financial context, watchlist, or research thesis. Use for read-only basket comparison and gap analysis without investment advice or live trading.","https:\u002F\u002Fgithub.com\u002FHoronomos-laps\u002FYana-AI\u002Fblob\u002Fmain\u002Fcore\u002Fskills\u002Fito-basket-compare\u002FSKILL.md",[],[],{"skillNo":65,"name":66,"nameZh":6,"description":67,"descriptionZh":6,"repositoryFullName":8,"ownerLogin":9,"ownerAvatarUrl":10,"ownerAccountNo":11,"ownerDisplayName":12,"repositoryNo":13,"starsCount":14,"licenseSpdxId":15,"licenseFileUrl":68,"sourceHtmlUrl":69,"categories":70,"keywords":71,"publishedAt":19},"01M734CYEMZFDR3KH6SW0JP8TD","implementing-network-traffic-baselining","Build network traffic baselines from NetFlow\u002FIPFIX data using Python pandas for statistical analysis, z-score anomaly detection, and hourly\u002Fdaily traffic pattern profiling","https:\u002F\u002Fgithub.com\u002Fyanacuti1121\u002FYana-AI\u002Fblob\u002Fmain\u002F.claude\u002Fskills\u002Fimplementing-network-traffic-baselining\u002FLICENSE","https:\u002F\u002Fgithub.com\u002FHoronomos-laps\u002FYana-AI\u002Fblob\u002Fmain\u002F.claude\u002Fskills\u002Fimplementing-network-traffic-baselining\u002FSKILL.md",[],[],{"skillNo":73,"name":74,"nameZh":6,"description":75,"descriptionZh":6,"repositoryFullName":8,"ownerLogin":9,"ownerAvatarUrl":10,"ownerAccountNo":11,"ownerDisplayName":12,"repositoryNo":13,"starsCount":14,"licenseSpdxId":15,"licenseFileUrl":76,"sourceHtmlUrl":77,"categories":78,"keywords":79,"publishedAt":19},"01M73ASZ6NJRJM44Z2P84NGQ3Q","detecting-evasion-techniques-in-endpoint-logs","Detects defense evasion techniques used by adversaries in endpoint logs including log tampering, timestomping, process injection, and security tool disabling. Use when investigating suspicious endpoint behavior, building detection rules for evasion tactics, or conducting threat hunting for stealthy adversary activity. Activates for requests involving evasion detection, defense evasion analysis, log tampering detection, or MITRE ATT&CK TA0005.","https:\u002F\u002Fgithub.com\u002Fyanacuti1121\u002FYana-AI\u002Fblob\u002Fmain\u002Fcore\u002Fskills\u002Fdetecting-evasion-techniques-in-endpoint-logs\u002FLICENSE","https:\u002F\u002Fgithub.com\u002FHoronomos-laps\u002FYana-AI\u002Fblob\u002Fmain\u002Fcore\u002Fskills\u002Fdetecting-evasion-techniques-in-endpoint-logs\u002FSKILL.md",[],[],{"skillNo":81,"name":82,"nameZh":6,"description":83,"descriptionZh":6,"repositoryFullName":8,"ownerLogin":9,"ownerAvatarUrl":10,"ownerAccountNo":11,"ownerDisplayName":12,"repositoryNo":13,"starsCount":14,"licenseSpdxId":15,"licenseFileUrl":84,"sourceHtmlUrl":85,"categories":86,"keywords":87,"publishedAt":19},"01M734AFNXD06Z78QRK0VKGT7Q","implementing-microsegmentation-with-guardicore","Implementing microsegmentation using Akamai Guardicore Segmentation to map application dependencies, create granular network policies, visualize east-west traffic flows, and enforce least-privilege communication between workloads across data centers and cloud.","https:\u002F\u002Fgithub.com\u002Fyanacuti1121\u002FYana-AI\u002Fblob\u002Fmain\u002F.claude\u002Fskills\u002Fimplementing-microsegmentation-with-guardicore\u002FLICENSE","https:\u002F\u002Fgithub.com\u002FHoronomos-laps\u002FYana-AI\u002Fblob\u002Fmain\u002F.claude\u002Fskills\u002Fimplementing-microsegmentation-with-guardicore\u002FSKILL.md",[],[],{"skillNo":89,"name":90,"nameZh":6,"description":6,"descriptionZh":6,"repositoryFullName":8,"ownerLogin":9,"ownerAvatarUrl":10,"ownerAccountNo":11,"ownerDisplayName":12,"repositoryNo":13,"starsCount":14,"licenseSpdxId":15,"licenseFileUrl":6,"sourceHtmlUrl":91,"categories":92,"keywords":93,"publishedAt":19},"01M736TK27QMCXK0DVCH25SJN5","ramdisk-tmpfs-patterns","https:\u002F\u002Fgithub.com\u002FHoronomos-laps\u002FYana-AI\u002Fblob\u002Fmain\u002F.claude\u002Fskills\u002Framdisk-tmpfs-patterns\u002FSKILL.md",[],[]]