{"title":"Skills de Agents de AI y operaciones de LLM","description":"\u003cp\u003eSkills de IA de nivel de producción para las personas que construyen y gestionan agentes LLM: RAG y recuperación, ingeniería de contexto, orquestación de agentes, evaluación, barreras de seguridad, LLMOps e inferencia, ajuste fino, voz y en tiempo real, costo y latencia, memoria y grafos de conocimiento, y control de calidad de agentes. Cada Skill convierte a Claude, ChatGPT y otros modelos en un ingeniero senior de IA que entrega artefactos reales y revisables.\u003c\/p\u003e","products":[{"product_id":"kai-rag-retrieval-architect","title":"Kai — RAG \/ Retrieval Architect AI Skill","description":"\u003cdiv style=\"font-family: 'DM Sans', sans-serif; color: #1A1A18; max-width: 680px;\"\u003e\n  \u003cp style=\"font-size: 16px; font-weight: 600; line-height: 1.5; margin: 0 0 8px 0;\"\u003eDrop Kai into Claude and get a senior RAG architect who builds retrieval that returns the right context, measured with an eval harness before you tune.\u003c\/p\u003e\n  \u003cp style=\"font-size: 13px; color: #555550; line-height: 1.7; margin: 0 0 28px 0;\"\u003eKai treats retrieval as an information-retrieval problem first: corpus prep and chunking strategy, embedding selection, vector index tuning (HNSW\/IVF), hybrid search (BM25 + dense) with RRF fusion, cross-encoder reranking, query transformation (multi-query, HyDE, decomposition), retrieval and end-to-end evaluation (recall@k, MRR, nDCG, faithfulness), grounding and citation, freshness, and the cost and latency of every hop. Measure before you tune. Always evaluate on a held-out labeled set in dev\/staging before serving in production.\u003c\/p\u003e\n  \u003cdiv style=\"background: #ECEDFC; border-radius: 12px; padding: 24px 28px; margin-bottom: 24px;\"\u003e\n    \u003cp style=\"font-size: 10px; font-weight: 600; color: #4A5BEE; letter-spacing: 0.08em; text-transform: uppercase; margin: 0 0 16px 0;\"\u003eWhat you get\u003c\/p\u003e\n    \u003cul style=\"margin: 0; padding: 0; list-style: none;\"\u003e\n\u003cli style=\"font-size: 13px; padding: 7px 0; border-bottom: 1px solid rgba(74,91,238,0.14); display: flex; gap: 10px;\"\u003e\n\u003cspan style=\"color:#4A5BEE; font-weight:600;\"\u003e→\u003c\/span\u003e\u003cspan\u003eChunking strategy and metadata design\u003c\/span\u003e\n\u003c\/li\u003e\n\u003cli style=\"font-size: 13px; padding: 7px 0; border-bottom: 1px solid rgba(74,91,238,0.14); display: flex; gap: 10px;\"\u003e\n\u003cspan style=\"color:#4A5BEE; font-weight:600;\"\u003e→\u003c\/span\u003e\u003cspan\u003eHybrid search (BM25 + dense) with RRF and cross-encoder reranking\u003c\/span\u003e\n\u003c\/li\u003e\n\u003cli style=\"font-size: 13px; padding: 7px 0; border-bottom: 1px solid rgba(74,91,238,0.14); display: flex; gap: 10px;\"\u003e\n\u003cspan style=\"color:#4A5BEE; font-weight:600;\"\u003e→\u003c\/span\u003e\u003cspan\u003eQuery transformation: multi-query, HyDE, decomposition\u003c\/span\u003e\n\u003c\/li\u003e\n\u003cli style=\"font-size: 13px; padding: 7px 0;  display: flex; gap: 10px;\"\u003e\n\u003cspan style=\"color:#4A5BEE; font-weight:600;\"\u003e→\u003c\/span\u003e\u003cspan\u003eRetrieval eval: recall@k, MRR, nDCG, faithfulness, with before\/after\u003c\/span\u003e\n\u003c\/li\u003e\n    \u003c\/ul\u003e\n  \u003c\/div\u003e\n  \u003cdiv style=\"display:flex; align-items:center; gap:20px; background:#FFFFFF; border:1px solid #E8E6E0; border-radius:8px; padding:14px 20px; margin-bottom:24px;\"\u003e\n    \u003cspan style=\"font-size:11px; color:#888780; font-family:monospace;\"\u003e📄 kai-rag-retrieval-architect.skill\u003c\/span\u003e\n    \u003cspan style=\"font-size:11px; color:#888780;\"\u003eUnder 2 min install\u003c\/span\u003e\n    \u003cspan style=\"font-size:11px; color:#888780;\"\u003eWorks with Claude, ChatGPT \u0026amp; any AI chat\u003c\/span\u003e\n  \u003c\/div\u003e\n  \u003cdiv style=\"border-left:3px solid #4A5BEE; padding-left:16px;\"\u003e\n    \u003cp style=\"font-size:10px; font-weight:600; color:#4A5BEE; letter-spacing:0.08em; text-transform:uppercase; margin:0 0 6px 0;\"\u003eHow to install\u003c\/p\u003e\n    \u003cp style=\"font-size:12px; color:#555550; line-height:1.7; margin:0;\"\u003eDownload the .skill package → open Claude → paste SKILL.md into your Project Instructions or system prompt → describe your requirement → Kai builds the answer. Includes a full worked example so you see exactly what you get.\u003c\/p\u003e\n  \u003c\/div\u003e\n\u003c\/div\u003e","brand":"KissMySkills","offers":[{"title":"Default Title","offer_id":58312988721416,"sku":null,"price":34.0,"currency_code":"USD","in_stock":true}],"thumbnail_url":"\/\/cdn.shopify.com\/s\/files\/1\/1036\/1444\/7880\/files\/ag2-kai-cover.png?v=1785164122"},{"product_id":"selin-context-engineer","title":"Selin — Context Engineer AI Skill","description":"\u003cdiv style=\"font-family: 'DM Sans', sans-serif; color: #1A1A18; max-width: 680px;\"\u003e\n  \u003cp style=\"font-size: 16px; font-weight: 600; line-height: 1.5; margin: 0 0 8px 0;\"\u003eDrop Selin into Claude and get a senior context engineer who treats the context window as a scarce, ordered token budget and makes every token earn its place.\u003c\/p\u003e\n  \u003cp style=\"font-size: 13px; color: #555550; line-height: 1.7; margin: 0 0 28px 0;\"\u003eSelin engineers the context window: token accounting across system prompt, tools, retrieved snippets, memory and history; context assembly and templating; positional placement (lost-in-the-middle, primacy and recency); compression and summarization before truncation; retrieved-context selection and dedup; tool-result trimming; memory injection; multi-turn history windowing and eviction; structured context; and prompt caching. Measure whether the model actually uses the context. Always evaluate on a held-out set in dev\/staging before production.\u003c\/p\u003e\n  \u003cdiv style=\"background: #ECEDFC; border-radius: 12px; padding: 24px 28px; margin-bottom: 24px;\"\u003e\n    \u003cp style=\"font-size: 10px; font-weight: 600; color: #4A5BEE; letter-spacing: 0.08em; text-transform: uppercase; margin: 0 0 16px 0;\"\u003eWhat you get\u003c\/p\u003e\n    \u003cul style=\"margin: 0; padding: 0; list-style: none;\"\u003e\n\u003cli style=\"font-size: 13px; padding: 7px 0; border-bottom: 1px solid rgba(74,91,238,0.14); display: flex; gap: 10px;\"\u003e\n\u003cspan style=\"color:#4A5BEE; font-weight:600;\"\u003e→\u003c\/span\u003e\u003cspan\u003eToken accounting and per-region context budgets\u003c\/span\u003e\n\u003c\/li\u003e\n\u003cli style=\"font-size: 13px; padding: 7px 0; border-bottom: 1px solid rgba(74,91,238,0.14); display: flex; gap: 10px;\"\u003e\n\u003cspan style=\"color:#4A5BEE; font-weight:600;\"\u003e→\u003c\/span\u003e\u003cspan\u003ePlacement for lost-in-the-middle, primacy and recency\u003c\/span\u003e\n\u003c\/li\u003e\n\u003cli style=\"font-size: 13px; padding: 7px 0; border-bottom: 1px solid rgba(74,91,238,0.14); display: flex; gap: 10px;\"\u003e\n\u003cspan style=\"color:#4A5BEE; font-weight:600;\"\u003e→\u003c\/span\u003e\u003cspan\u003eCompression, dedup and tool-result trimming before truncation\u003c\/span\u003e\n\u003c\/li\u003e\n\u003cli style=\"font-size: 13px; padding: 7px 0;  display: flex; gap: 10px;\"\u003e\n\u003cspan style=\"color:#4A5BEE; font-weight:600;\"\u003e→\u003c\/span\u003e\u003cspan\u003eHistory windowing, memory injection and prompt caching\u003c\/span\u003e\n\u003c\/li\u003e\n    \u003c\/ul\u003e\n  \u003c\/div\u003e\n  \u003cdiv style=\"display:flex; align-items:center; gap:20px; background:#FFFFFF; border:1px solid #E8E6E0; border-radius:8px; padding:14px 20px; margin-bottom:24px;\"\u003e\n    \u003cspan style=\"font-size:11px; color:#888780; font-family:monospace;\"\u003e📄 selin-context-engineer.skill\u003c\/span\u003e\n    \u003cspan style=\"font-size:11px; color:#888780;\"\u003eUnder 2 min install\u003c\/span\u003e\n    \u003cspan style=\"font-size:11px; color:#888780;\"\u003eWorks with Claude, ChatGPT \u0026amp; any AI chat\u003c\/span\u003e\n  \u003c\/div\u003e\n  \u003cdiv style=\"border-left:3px solid #4A5BEE; padding-left:16px;\"\u003e\n    \u003cp style=\"font-size:10px; font-weight:600; color:#4A5BEE; letter-spacing:0.08em; text-transform:uppercase; margin:0 0 6px 0;\"\u003eHow to install\u003c\/p\u003e\n    \u003cp style=\"font-size:12px; color:#555550; line-height:1.7; margin:0;\"\u003eDownload the .skill package → open Claude → paste SKILL.md into your Project Instructions or system prompt → describe your requirement → Selin builds the answer. Includes a full worked example so you see exactly what you get.\u003c\/p\u003e\n  \u003c\/div\u003e\n\u003c\/div\u003e","brand":"KissMySkills","offers":[{"title":"Default Title","offer_id":58312991899912,"sku":null,"price":34.0,"currency_code":"USD","in_stock":true}],"thumbnail_url":"\/\/cdn.shopify.com\/s\/files\/1\/1036\/1444\/7880\/files\/ag2-selin-cover.png?v=1785164182"},{"product_id":"enzo-ai-agent-orchestrator","title":"Enzo — AI Agent Orchestrator AI Skill","description":"\u003cdiv style=\"font-family: 'DM Sans', sans-serif; color: #1A1A18; max-width: 680px;\"\u003e\n  \u003cp style=\"font-size: 16px; font-weight: 600; line-height: 1.5; margin: 0 0 8px 0;\"\u003eDrop Enzo into Claude and get a senior agent orchestrator who builds the agent loop: typed state, validated tool calls, retries, loop guards and traces.\u003c\/p\u003e\n  \u003cp style=\"font-size: 13px; color: #555550; line-height: 1.7; margin: 0 0 28px 0;\"\u003eEnzo builds agent runtimes: the perceive-plan-act-observe loop, tool and function calling with JSON schemas, argument validation, timeout\/retry and idempotency, planning patterns (ReAct, plan-and-execute, reflection), multi-agent topologies (supervisor\/worker, hierarchical, sequential) and when each earns its cost, typed state and checkpoints, human-in-the-loop, loop-guarding and budgets, observability and tracing, and the cost and latency of multi-step runs. Framework-aware (LangGraph, CrewAI, AutoGen, OpenAI Agents SDK, Anthropic tool use), not married to one. Always test in staging with guardrails, budgets and traces before production.\u003c\/p\u003e\n  \u003cdiv style=\"background: #ECEDFC; border-radius: 12px; padding: 24px 28px; margin-bottom: 24px;\"\u003e\n    \u003cp style=\"font-size: 10px; font-weight: 600; color: #4A5BEE; letter-spacing: 0.08em; text-transform: uppercase; margin: 0 0 16px 0;\"\u003eWhat you get\u003c\/p\u003e\n    \u003cul style=\"margin: 0; padding: 0; list-style: none;\"\u003e\n\u003cli style=\"font-size: 13px; padding: 7px 0; border-bottom: 1px solid rgba(74,91,238,0.14); display: flex; gap: 10px;\"\u003e\n\u003cspan style=\"color:#4A5BEE; font-weight:600;\"\u003e→\u003c\/span\u003e\u003cspan\u003eThe agent loop with schema-validated tool calls\u003c\/span\u003e\n\u003c\/li\u003e\n\u003cli style=\"font-size: 13px; padding: 7px 0; border-bottom: 1px solid rgba(74,91,238,0.14); display: flex; gap: 10px;\"\u003e\n\u003cspan style=\"color:#4A5BEE; font-weight:600;\"\u003e→\u003c\/span\u003e\u003cspan\u003eRetries, timeouts, idempotency and loop guards\u003c\/span\u003e\n\u003c\/li\u003e\n\u003cli style=\"font-size: 13px; padding: 7px 0; border-bottom: 1px solid rgba(74,91,238,0.14); display: flex; gap: 10px;\"\u003e\n\u003cspan style=\"color:#4A5BEE; font-weight:600;\"\u003e→\u003c\/span\u003e\u003cspan\u003ePlanning patterns and multi-agent topologies (with cost\/benefit)\u003c\/span\u003e\n\u003c\/li\u003e\n\u003cli style=\"font-size: 13px; padding: 7px 0;  display: flex; gap: 10px;\"\u003e\n\u003cspan style=\"color:#4A5BEE; font-weight:600;\"\u003e→\u003c\/span\u003e\u003cspan\u003eTyped state, human-in-the-loop, tracing and cost\/latency\u003c\/span\u003e\n\u003c\/li\u003e\n    \u003c\/ul\u003e\n  \u003c\/div\u003e\n  \u003cdiv style=\"display:flex; align-items:center; gap:20px; background:#FFFFFF; border:1px solid #E8E6E0; border-radius:8px; padding:14px 20px; margin-bottom:24px;\"\u003e\n    \u003cspan style=\"font-size:11px; color:#888780; font-family:monospace;\"\u003e📄 enzo-ai-agent-orchestrator.skill\u003c\/span\u003e\n    \u003cspan style=\"font-size:11px; color:#888780;\"\u003eUnder 2 min install\u003c\/span\u003e\n    \u003cspan style=\"font-size:11px; color:#888780;\"\u003eWorks with Claude, ChatGPT \u0026amp; any AI chat\u003c\/span\u003e\n  \u003c\/div\u003e\n  \u003cdiv style=\"border-left:3px solid #4A5BEE; padding-left:16px;\"\u003e\n    \u003cp style=\"font-size:10px; font-weight:600; color:#4A5BEE; letter-spacing:0.08em; text-transform:uppercase; margin:0 0 6px 0;\"\u003eHow to install\u003c\/p\u003e\n    \u003cp style=\"font-size:12px; color:#555550; line-height:1.7; margin:0;\"\u003eDownload the .skill package → open Claude → paste SKILL.md into your Project Instructions or system prompt → describe your requirement → Enzo builds the answer. Includes a full worked example so you see exactly what you get.\u003c\/p\u003e\n  \u003c\/div\u003e\n\u003c\/div\u003e","brand":"KissMySkills","offers":[{"title":"Default Title","offer_id":58312997404936,"sku":null,"price":39.0,"currency_code":"USD","in_stock":true}],"thumbnail_url":"\/\/cdn.shopify.com\/s\/files\/1\/1036\/1444\/7880\/files\/ag2-enzo-cover.png?v=1785164229"},{"product_id":"ingrid-llm-evaluation-engineer","title":"Ingrid — LLM Evaluation Engineer AI Skill","description":"\u003cdiv style=\"font-family: 'DM Sans', sans-serif; color: #1A1A18; max-width: 680px;\"\u003e\n  \u003cp style=\"font-size: 16px; font-weight: 600; line-height: 1.5; margin: 0 0 8px 0;\"\u003eDrop Ingrid into Claude and get a senior LLM evaluation engineer whose rule is simple: no ship without an eval gate.\u003c\/p\u003e\n  \u003cp style=\"font-size: 13px; color: #555550; line-height: 1.7; margin: 0 0 28px 0;\"\u003eIngrid evaluates LLM apps and agents: building golden datasets (coverage, edge cases, no leakage, versioning), metric selection, LLM-as-judge (rubric design, pairwise vs pointwise, bias and position effects with mitigations, calibration against human labels), RAG metrics (faithfulness, context and answer relevance), agent trajectory eval (tool-call correctness, task completion), offline harnesses and CI regression gates, online eval (A\/B, guardrail metrics, human review sampling), and statistical rigor (confidence intervals, significance). Tools like Ragas, promptfoo, DeepEval, LangSmith and Braintrust, without lock-in. Always evaluate on a held-out set before shipping.\u003c\/p\u003e\n  \u003cdiv style=\"background: #ECEDFC; border-radius: 12px; padding: 24px 28px; margin-bottom: 24px;\"\u003e\n    \u003cp style=\"font-size: 10px; font-weight: 600; color: #4A5BEE; letter-spacing: 0.08em; text-transform: uppercase; margin: 0 0 16px 0;\"\u003eWhat you get\u003c\/p\u003e\n    \u003cul style=\"margin: 0; padding: 0; list-style: none;\"\u003e\n\u003cli style=\"font-size: 13px; padding: 7px 0; border-bottom: 1px solid rgba(74,91,238,0.14); display: flex; gap: 10px;\"\u003e\n\u003cspan style=\"color:#4A5BEE; font-weight:600;\"\u003e→\u003c\/span\u003e\u003cspan\u003eGolden datasets: coverage, edge cases, no leakage\u003c\/span\u003e\n\u003c\/li\u003e\n\u003cli style=\"font-size: 13px; padding: 7px 0; border-bottom: 1px solid rgba(74,91,238,0.14); display: flex; gap: 10px;\"\u003e\n\u003cspan style=\"color:#4A5BEE; font-weight:600;\"\u003e→\u003c\/span\u003e\u003cspan\u003eLLM-as-judge: rubric, bias mitigation, human calibration\u003c\/span\u003e\n\u003c\/li\u003e\n\u003cli style=\"font-size: 13px; padding: 7px 0; border-bottom: 1px solid rgba(74,91,238,0.14); display: flex; gap: 10px;\"\u003e\n\u003cspan style=\"color:#4A5BEE; font-weight:600;\"\u003e→\u003c\/span\u003e\u003cspan\u003eRAG and agent metrics (faithfulness, tool-call accuracy)\u003c\/span\u003e\n\u003c\/li\u003e\n\u003cli style=\"font-size: 13px; padding: 7px 0;  display: flex; gap: 10px;\"\u003e\n\u003cspan style=\"color:#4A5BEE; font-weight:600;\"\u003e→\u003c\/span\u003e\u003cspan\u003eCI regression gates with confidence intervals and significance\u003c\/span\u003e\n\u003c\/li\u003e\n    \u003c\/ul\u003e\n  \u003c\/div\u003e\n  \u003cdiv style=\"display:flex; align-items:center; gap:20px; background:#FFFFFF; border:1px solid #E8E6E0; border-radius:8px; padding:14px 20px; margin-bottom:24px;\"\u003e\n    \u003cspan style=\"font-size:11px; color:#888780; font-family:monospace;\"\u003e📄 ingrid-llm-evaluation-engineer.skill\u003c\/span\u003e\n    \u003cspan style=\"font-size:11px; color:#888780;\"\u003eUnder 2 min install\u003c\/span\u003e\n    \u003cspan style=\"font-size:11px; color:#888780;\"\u003eWorks with Claude, ChatGPT \u0026amp; any AI chat\u003c\/span\u003e\n  \u003c\/div\u003e\n  \u003cdiv style=\"border-left:3px solid #4A5BEE; padding-left:16px;\"\u003e\n    \u003cp style=\"font-size:10px; font-weight:600; color:#4A5BEE; letter-spacing:0.08em; text-transform:uppercase; margin:0 0 6px 0;\"\u003eHow to install\u003c\/p\u003e\n    \u003cp style=\"font-size:12px; color:#555550; line-height:1.7; margin:0;\"\u003eDownload the .skill package → open Claude → paste SKILL.md into your Project Instructions or system prompt → describe your requirement → Ingrid builds the answer. Includes a full worked example so you see exactly what you get.\u003c\/p\u003e\n  \u003c\/div\u003e\n\u003c\/div\u003e","brand":"KissMySkills","offers":[{"title":"Default Title","offer_id":58313000845576,"sku":null,"price":34.0,"currency_code":"USD","in_stock":true}],"thumbnail_url":"\/\/cdn.shopify.com\/s\/files\/1\/1036\/1444\/7880\/files\/ag2-ingrid-cover.png?v=1785164278"},{"product_id":"rania-ai-guardrails-engineer","title":"Rania — AI Guardrails \u0026 Safety Engineer AI Skill","description":"\u003cdiv style=\"font-family: 'DM Sans', sans-serif; color: #1A1A18; max-width: 680px;\"\u003e\n  \u003cp style=\"font-size: 16px; font-weight: 600; line-height: 1.5; margin: 0 0 8px 0;\"\u003eDrop Rania into Claude and get a senior guardrails engineer who hardens your own agent against prompt injection and data leakage, defense only.\u003c\/p\u003e\n  \u003cp style=\"font-size: 13px; color: #555550; line-height: 1.7; margin: 0 0 28px 0;\"\u003eRania builds defensive guardrails for your LLM app: input guardrails (prompt-injection and jailbreak detection and mitigation, instruction-hierarchy enforcement, delimiting untrusted content), output guardrails (schema and format validation, PII detection and redaction, moderation, faithfulness and leakage checks), tool-use guardrails (least privilege, argument validation, confirmation gates for high-impact actions), rate and abuse limiting, and layered defense measured with your own red-team suite. Tools like Llama Guard, NeMo Guardrails, Guardrails AI and Presidio, without lock-in. Strictly defensive: no working jailbreaks, exploits or malware. Test on a held-out set in staging before production.\u003c\/p\u003e\n  \u003cdiv style=\"background: #ECEDFC; border-radius: 12px; padding: 24px 28px; margin-bottom: 24px;\"\u003e\n    \u003cp style=\"font-size: 10px; font-weight: 600; color: #4A5BEE; letter-spacing: 0.08em; text-transform: uppercase; margin: 0 0 16px 0;\"\u003eWhat you get\u003c\/p\u003e\n    \u003cul style=\"margin: 0; padding: 0; list-style: none;\"\u003e\n\u003cli style=\"font-size: 13px; padding: 7px 0; border-bottom: 1px solid rgba(74,91,238,0.14); display: flex; gap: 10px;\"\u003e\n\u003cspan style=\"color:#4A5BEE; font-weight:600;\"\u003e→\u003c\/span\u003e\u003cspan\u003ePrompt-injection and jailbreak detection and mitigation (defensive)\u003c\/span\u003e\n\u003c\/li\u003e\n\u003cli style=\"font-size: 13px; padding: 7px 0; border-bottom: 1px solid rgba(74,91,238,0.14); display: flex; gap: 10px;\"\u003e\n\u003cspan style=\"color:#4A5BEE; font-weight:600;\"\u003e→\u003c\/span\u003e\u003cspan\u003eOutput guardrails: schema validation, PII redaction, moderation\u003c\/span\u003e\n\u003c\/li\u003e\n\u003cli style=\"font-size: 13px; padding: 7px 0; border-bottom: 1px solid rgba(74,91,238,0.14); display: flex; gap: 10px;\"\u003e\n\u003cspan style=\"color:#4A5BEE; font-weight:600;\"\u003e→\u003c\/span\u003e\u003cspan\u003eTool guardrails: least privilege, arg validation, confirmation gates\u003c\/span\u003e\n\u003c\/li\u003e\n\u003cli style=\"font-size: 13px; padding: 7px 0;  display: flex; gap: 10px;\"\u003e\n\u003cspan style=\"color:#4A5BEE; font-weight:600;\"\u003e→\u003c\/span\u003e\u003cspan\u003eLayered defense measured with your own red-team suite\u003c\/span\u003e\n\u003c\/li\u003e\n    \u003c\/ul\u003e\n  \u003c\/div\u003e\n  \u003cdiv style=\"display:flex; align-items:center; gap:20px; background:#FFFFFF; border:1px solid #E8E6E0; border-radius:8px; padding:14px 20px; margin-bottom:24px;\"\u003e\n    \u003cspan style=\"font-size:11px; color:#888780; font-family:monospace;\"\u003e📄 rania-ai-guardrails-engineer.skill\u003c\/span\u003e\n    \u003cspan style=\"font-size:11px; color:#888780;\"\u003eUnder 2 min install\u003c\/span\u003e\n    \u003cspan style=\"font-size:11px; color:#888780;\"\u003eWorks with Claude, ChatGPT \u0026amp; any AI chat\u003c\/span\u003e\n  \u003c\/div\u003e\n  \u003cdiv style=\"border-left:3px solid #4A5BEE; padding-left:16px;\"\u003e\n    \u003cp style=\"font-size:10px; font-weight:600; color:#4A5BEE; letter-spacing:0.08em; text-transform:uppercase; margin:0 0 6px 0;\"\u003eHow to install\u003c\/p\u003e\n    \u003cp style=\"font-size:12px; color:#555550; line-height:1.7; margin:0;\"\u003eDownload the .skill package → open Claude → paste SKILL.md into your Project Instructions or system prompt → describe your requirement → Rania builds the answer. Includes a full worked example so you see exactly what you get.\u003c\/p\u003e\n  \u003c\/div\u003e\n\u003c\/div\u003e","brand":"KissMySkills","offers":[{"title":"Default Title","offer_id":58313004581128,"sku":null,"price":34.0,"currency_code":"USD","in_stock":true}],"thumbnail_url":"\/\/cdn.shopify.com\/s\/files\/1\/1036\/1444\/7880\/files\/ag2-rania-cover.png?v=1785164314"},{"product_id":"dario-llmops-inference-engineer","title":"Dario — LLMOps \/ Inference Engineer AI Skill","description":"\u003cdiv style=\"font-family: 'DM Sans', sans-serif; color: #1A1A18; max-width: 680px;\"\u003e\n  \u003cp style=\"font-size: 16px; font-weight: 600; line-height: 1.5; margin: 0 0 8px 0;\"\u003eDrop Dario into Claude and get a senior LLMOps engineer who runs your model like an SRE-owned service: SLOs, a gateway, dashboards and one-step rollback.\u003c\/p\u003e\n  \u003cp style=\"font-size: 13px; color: #555550; line-height: 1.7; margin: 0 0 28px 0;\"\u003eDario serves and operates LLMs in production: inference engines (vLLM, TGI, TensorRT-LLM), throughput vs latency (continuous batching, KV-cache, paged attention, speculative decoding, quantization tradeoffs), an LLM gateway (routing, fallbacks, retries, timeouts, rate and quota limits, key management), autoscaling and GPU capacity planning, caching (prefix and semantic), reliability (circuit breakers, multi-provider failover), observability (p50\/p95\/p99, TTFT, tokens\/sec, cost per request, alerts), and canary\/shadow rollout with SLOs and error budgets. Tools like vLLM, LiteLLM, KServe, Prometheus, Grafana and Langfuse, without lock-in. Roll out via canary in staging first.\u003c\/p\u003e\n  \u003cdiv style=\"background: #ECEDFC; border-radius: 12px; padding: 24px 28px; margin-bottom: 24px;\"\u003e\n    \u003cp style=\"font-size: 10px; font-weight: 600; color: #4A5BEE; letter-spacing: 0.08em; text-transform: uppercase; margin: 0 0 16px 0;\"\u003eWhat you get\u003c\/p\u003e\n    \u003cul style=\"margin: 0; padding: 0; list-style: none;\"\u003e\n\u003cli style=\"font-size: 13px; padding: 7px 0; border-bottom: 1px solid rgba(74,91,238,0.14); display: flex; gap: 10px;\"\u003e\n\u003cspan style=\"color:#4A5BEE; font-weight:600;\"\u003e→\u003c\/span\u003e\u003cspan\u003eInference tuning: continuous batching, KV-cache, quantization\u003c\/span\u003e\n\u003c\/li\u003e\n\u003cli style=\"font-size: 13px; padding: 7px 0; border-bottom: 1px solid rgba(74,91,238,0.14); display: flex; gap: 10px;\"\u003e\n\u003cspan style=\"color:#4A5BEE; font-weight:600;\"\u003e→\u003c\/span\u003e\u003cspan\u003eAn LLM gateway: routing, fallbacks, retries, rate limits\u003c\/span\u003e\n\u003c\/li\u003e\n\u003cli style=\"font-size: 13px; padding: 7px 0; border-bottom: 1px solid rgba(74,91,238,0.14); display: flex; gap: 10px;\"\u003e\n\u003cspan style=\"color:#4A5BEE; font-weight:600;\"\u003e→\u003c\/span\u003e\u003cspan\u003eCapacity math, autoscaling and prefix\/semantic caching\u003c\/span\u003e\n\u003c\/li\u003e\n\u003cli style=\"font-size: 13px; padding: 7px 0;  display: flex; gap: 10px;\"\u003e\n\u003cspan style=\"color:#4A5BEE; font-weight:600;\"\u003e→\u003c\/span\u003e\u003cspan\u003eObservability (TTFT, p95, cost\/req), SLOs and canary rollout\u003c\/span\u003e\n\u003c\/li\u003e\n    \u003c\/ul\u003e\n  \u003c\/div\u003e\n  \u003cdiv style=\"display:flex; align-items:center; gap:20px; background:#FFFFFF; border:1px solid #E8E6E0; border-radius:8px; padding:14px 20px; margin-bottom:24px;\"\u003e\n    \u003cspan style=\"font-size:11px; color:#888780; font-family:monospace;\"\u003e📄 dario-llmops-inference-engineer.skill\u003c\/span\u003e\n    \u003cspan style=\"font-size:11px; color:#888780;\"\u003eUnder 2 min install\u003c\/span\u003e\n    \u003cspan style=\"font-size:11px; color:#888780;\"\u003eWorks with Claude, ChatGPT \u0026amp; any AI chat\u003c\/span\u003e\n  \u003c\/div\u003e\n  \u003cdiv style=\"border-left:3px solid #4A5BEE; padding-left:16px;\"\u003e\n    \u003cp style=\"font-size:10px; font-weight:600; color:#4A5BEE; letter-spacing:0.08em; text-transform:uppercase; margin:0 0 6px 0;\"\u003eHow to install\u003c\/p\u003e\n    \u003cp style=\"font-size:12px; color:#555550; line-height:1.7; margin:0;\"\u003eDownload the .skill package → open Claude → paste SKILL.md into your Project Instructions or system prompt → describe your requirement → Dario builds the answer. Includes a full worked example so you see exactly what you get.\u003c\/p\u003e\n  \u003c\/div\u003e\n\u003c\/div\u003e","brand":"KissMySkills","offers":[{"title":"Default Title","offer_id":58313009463560,"sku":null,"price":34.0,"currency_code":"USD","in_stock":true}],"thumbnail_url":"\/\/cdn.shopify.com\/s\/files\/1\/1036\/1444\/7880\/files\/ag2-dario-cover.png?v=1785164357"},{"product_id":"mei-llm-finetuning-engineer","title":"Mei — LLM Fine-tuning Engineer AI Skill","description":"\u003cdiv style=\"font-family: 'DM Sans', sans-serif; color: #1A1A18; max-width: 680px;\"\u003e\n  \u003cp style=\"font-size: 16px; font-weight: 600; line-height: 1.5; margin: 0 0 8px 0;\"\u003eDrop Mei into Claude and get a senior fine-tuning engineer whose first question is: have you exhausted prompting and RAG? You earn the fine-tune and prove it with eval.\u003c\/p\u003e\n  \u003cp style=\"font-size: 13px; color: #555550; line-height: 1.7; margin: 0 0 28px 0;\"\u003eMei adapts LLMs the disciplined way: when not to fine-tune (prompt\/RAG first) and when it genuinely wins (format and schema adherence, tool-call reliability, latency and cost via smaller models, domain tone), dataset curation (quality over quantity, dedup, decontamination against the eval set, chat templating), methods (full SFT vs PEFT: LoRA\/QLoRA rank\/alpha\/target modules, learning rate, epochs, packing), preference tuning (DPO\/ORPO\/KTO), GPU-memory math, overfitting and catastrophic forgetting detection, evaluation vs the base model, and serving the adapter. Tools like Hugging Face TRL\/PEFT, Axolotl, Unsloth and bitsandbytes, without lock-in. Always evaluate on a held-out set before serving.\u003c\/p\u003e\n  \u003cdiv style=\"background: #ECEDFC; border-radius: 12px; padding: 24px 28px; margin-bottom: 24px;\"\u003e\n    \u003cp style=\"font-size: 10px; font-weight: 600; color: #4A5BEE; letter-spacing: 0.08em; text-transform: uppercase; margin: 0 0 16px 0;\"\u003eWhat you get\u003c\/p\u003e\n    \u003cul style=\"margin: 0; padding: 0; list-style: none;\"\u003e\n\u003cli style=\"font-size: 13px; padding: 7px 0; border-bottom: 1px solid rgba(74,91,238,0.14); display: flex; gap: 10px;\"\u003e\n\u003cspan style=\"color:#4A5BEE; font-weight:600;\"\u003e→\u003c\/span\u003e\u003cspan\u003eWhen NOT to fine-tune vs when it genuinely wins\u003c\/span\u003e\n\u003c\/li\u003e\n\u003cli style=\"font-size: 13px; padding: 7px 0; border-bottom: 1px solid rgba(74,91,238,0.14); display: flex; gap: 10px;\"\u003e\n\u003cspan style=\"color:#4A5BEE; font-weight:600;\"\u003e→\u003c\/span\u003e\u003cspan\u003eDataset curation: dedup, decontamination, chat templating\u003c\/span\u003e\n\u003c\/li\u003e\n\u003cli style=\"font-size: 13px; padding: 7px 0; border-bottom: 1px solid rgba(74,91,238,0.14); display: flex; gap: 10px;\"\u003e\n\u003cspan style=\"color:#4A5BEE; font-weight:600;\"\u003e→\u003c\/span\u003e\u003cspan\u003eQLoRA\/LoRA and preference tuning (DPO\/ORPO\/KTO) with defaults\u003c\/span\u003e\n\u003c\/li\u003e\n\u003cli style=\"font-size: 13px; padding: 7px 0;  display: flex; gap: 10px;\"\u003e\n\u003cspan style=\"color:#4A5BEE; font-weight:600;\"\u003e→\u003c\/span\u003e\u003cspan\u003eGPU-memory math, forgetting checks and base-vs-tuned eval\u003c\/span\u003e\n\u003c\/li\u003e\n    \u003c\/ul\u003e\n  \u003c\/div\u003e\n  \u003cdiv style=\"display:flex; align-items:center; gap:20px; background:#FFFFFF; border:1px solid #E8E6E0; border-radius:8px; padding:14px 20px; margin-bottom:24px;\"\u003e\n    \u003cspan style=\"font-size:11px; color:#888780; font-family:monospace;\"\u003e📄 mei-llm-finetuning-engineer.skill\u003c\/span\u003e\n    \u003cspan style=\"font-size:11px; color:#888780;\"\u003eUnder 2 min install\u003c\/span\u003e\n    \u003cspan style=\"font-size:11px; color:#888780;\"\u003eWorks with Claude, ChatGPT \u0026amp; any AI chat\u003c\/span\u003e\n  \u003c\/div\u003e\n  \u003cdiv style=\"border-left:3px solid #4A5BEE; padding-left:16px;\"\u003e\n    \u003cp style=\"font-size:10px; font-weight:600; color:#4A5BEE; letter-spacing:0.08em; text-transform:uppercase; margin:0 0 6px 0;\"\u003eHow to install\u003c\/p\u003e\n    \u003cp style=\"font-size:12px; color:#555550; line-height:1.7; margin:0;\"\u003eDownload the .skill package → open Claude → paste SKILL.md into your Project Instructions or system prompt → describe your requirement → Mei builds the answer. Includes a full worked example so you see exactly what you get.\u003c\/p\u003e\n  \u003c\/div\u003e\n\u003c\/div\u003e","brand":"KissMySkills","offers":[{"title":"Default Title","offer_id":58313013002504,"sku":null,"price":34.0,"currency_code":"USD","in_stock":true}],"thumbnail_url":"\/\/cdn.shopify.com\/s\/files\/1\/1036\/1444\/7880\/files\/ag2-mei-cover.png?v=1785164395"},{"product_id":"aksel-voice-realtime-agent-engineer","title":"Aksel — Voice \u0026 Realtime Agent Engineer AI Skill","description":"\u003cdiv style=\"font-family: 'DM Sans', sans-serif; color: #1A1A18; max-width: 680px;\"\u003e\n  \u003cp style=\"font-size: 16px; font-weight: 600; line-height: 1.5; margin: 0 0 8px 0;\"\u003eDrop Aksel into Claude and get a senior voice AI engineer who builds low-latency realtime agents where latency is the product.\u003c\/p\u003e\n  \u003cp style=\"font-size: 13px; color: #555550; line-height: 1.7; margin: 0 0 28px 0;\"\u003eAksel builds voice and realtime agents: the STT to LLM to TTS pipeline (and speech-to-speech realtime APIs), the latency budget (sub-800ms target broken down across VAD, endpointing, STT, LLM TTFT, TTS first-audio and network), streaming everything, turn-taking (voice activity detection, endpointing, barge-in and interruption), telephony vs web (SIP, WebRTC, codecs), mid-conversation tool calls, error recovery, and evaluation (latency percentiles, word error rate, task success). Tools like Deepgram, Whisper, ElevenLabs, Cartesia, LiveKit, Pipecat and the OpenAI Realtime API, without lock-in. Test in staging under real network conditions before production.\u003c\/p\u003e\n  \u003cdiv style=\"background: #ECEDFC; border-radius: 12px; padding: 24px 28px; margin-bottom: 24px;\"\u003e\n    \u003cp style=\"font-size: 10px; font-weight: 600; color: #4A5BEE; letter-spacing: 0.08em; text-transform: uppercase; margin: 0 0 16px 0;\"\u003eWhat you get\u003c\/p\u003e\n    \u003cul style=\"margin: 0; padding: 0; list-style: none;\"\u003e\n\u003cli style=\"font-size: 13px; padding: 7px 0; border-bottom: 1px solid rgba(74,91,238,0.14); display: flex; gap: 10px;\"\u003e\n\u003cspan style=\"color:#4A5BEE; font-weight:600;\"\u003e→\u003c\/span\u003e\u003cspan\u003eThe realtime STT-LLM-TTS pipeline and a per-stage latency budget\u003c\/span\u003e\n\u003c\/li\u003e\n\u003cli style=\"font-size: 13px; padding: 7px 0; border-bottom: 1px solid rgba(74,91,238,0.14); display: flex; gap: 10px;\"\u003e\n\u003cspan style=\"color:#4A5BEE; font-weight:600;\"\u003e→\u003c\/span\u003e\u003cspan\u003eTurn-taking: VAD, endpointing and barge-in handling\u003c\/span\u003e\n\u003c\/li\u003e\n\u003cli style=\"font-size: 13px; padding: 7px 0; border-bottom: 1px solid rgba(74,91,238,0.14); display: flex; gap: 10px;\"\u003e\n\u003cspan style=\"color:#4A5BEE; font-weight:600;\"\u003e→\u003c\/span\u003e\u003cspan\u003eTelephony vs web (SIP, WebRTC) and mid-call tool calls\u003c\/span\u003e\n\u003c\/li\u003e\n\u003cli style=\"font-size: 13px; padding: 7px 0;  display: flex; gap: 10px;\"\u003e\n\u003cspan style=\"color:#4A5BEE; font-weight:600;\"\u003e→\u003c\/span\u003e\u003cspan\u003eEvaluation: latency percentiles, word error rate, task success\u003c\/span\u003e\n\u003c\/li\u003e\n    \u003c\/ul\u003e\n  \u003c\/div\u003e\n  \u003cdiv style=\"display:flex; align-items:center; gap:20px; background:#FFFFFF; border:1px solid #E8E6E0; border-radius:8px; padding:14px 20px; margin-bottom:24px;\"\u003e\n    \u003cspan style=\"font-size:11px; color:#888780; font-family:monospace;\"\u003e📄 aksel-voice-realtime-agent-engineer.skill\u003c\/span\u003e\n    \u003cspan style=\"font-size:11px; color:#888780;\"\u003eUnder 2 min install\u003c\/span\u003e\n    \u003cspan style=\"font-size:11px; color:#888780;\"\u003eWorks with Claude, ChatGPT \u0026amp; any AI chat\u003c\/span\u003e\n  \u003c\/div\u003e\n  \u003cdiv style=\"border-left:3px solid #4A5BEE; padding-left:16px;\"\u003e\n    \u003cp style=\"font-size:10px; font-weight:600; color:#4A5BEE; letter-spacing:0.08em; text-transform:uppercase; margin:0 0 6px 0;\"\u003eHow to install\u003c\/p\u003e\n    \u003cp style=\"font-size:12px; color:#555550; line-height:1.7; margin:0;\"\u003eDownload the .skill package → open Claude → paste SKILL.md into your Project Instructions or system prompt → describe your requirement → Aksel builds the answer. Includes a full worked example so you see exactly what you get.\u003c\/p\u003e\n  \u003c\/div\u003e\n\u003c\/div\u003e","brand":"KissMySkills","offers":[{"title":"Default Title","offer_id":58313018474760,"sku":null,"price":34.0,"currency_code":"USD","in_stock":true}],"thumbnail_url":"\/\/cdn.shopify.com\/s\/files\/1\/1036\/1444\/7880\/files\/ag2-aksel-cover.png?v=1785164432"},{"product_id":"devika-llm-cost-latency-optimizer","title":"Devika — LLM Cost \u0026 Latency Optimizer AI Skill","description":"\u003cdiv style=\"font-family: 'DM Sans', sans-serif; color: #1A1A18; max-width: 680px;\"\u003e\n  \u003cp style=\"font-size: 16px; font-weight: 600; line-height: 1.5; margin: 0 0 8px 0;\"\u003eDrop Devika into Claude and get a senior cost and latency optimizer who cuts spend and tail latency while measuring quality on every change.\u003c\/p\u003e\n  \u003cp style=\"font-size: 13px; color: #555550; line-height: 1.7; margin: 0 0 28px 0;\"\u003eDevika cuts the cost and latency of LLM features without wrecking quality: token accounting and cost modeling, model routing and cascades (cheap model first, escalate on low confidence), prompt slimming (shorter prompts, fewer and better few-shots, output-length control, structured output), caching (exact, prefix and semantic with thresholds), batching, retrieval trimming, streaming for perceived latency, and quantization tradeoffs, always with a quality gate so you never trade cost for silent quality loss. Tools like LiteLLM, Helicone, Langfuse and tiktoken, without lock-in. Measure quality on a held-out set for every cost cut.\u003c\/p\u003e\n  \u003cdiv style=\"background: #ECEDFC; border-radius: 12px; padding: 24px 28px; margin-bottom: 24px;\"\u003e\n    \u003cp style=\"font-size: 10px; font-weight: 600; color: #4A5BEE; letter-spacing: 0.08em; text-transform: uppercase; margin: 0 0 16px 0;\"\u003eWhat you get\u003c\/p\u003e\n    \u003cul style=\"margin: 0; padding: 0; list-style: none;\"\u003e\n\u003cli style=\"font-size: 13px; padding: 7px 0; border-bottom: 1px solid rgba(74,91,238,0.14); display: flex; gap: 10px;\"\u003e\n\u003cspan style=\"color:#4A5BEE; font-weight:600;\"\u003e→\u003c\/span\u003e\u003cspan\u003eToken accounting and a cost model that finds the big line items\u003c\/span\u003e\n\u003c\/li\u003e\n\u003cli style=\"font-size: 13px; padding: 7px 0; border-bottom: 1px solid rgba(74,91,238,0.14); display: flex; gap: 10px;\"\u003e\n\u003cspan style=\"color:#4A5BEE; font-weight:600;\"\u003e→\u003c\/span\u003e\u003cspan\u003eModel routing\/cascades and prompt slimming\u003c\/span\u003e\n\u003c\/li\u003e\n\u003cli style=\"font-size: 13px; padding: 7px 0; border-bottom: 1px solid rgba(74,91,238,0.14); display: flex; gap: 10px;\"\u003e\n\u003cspan style=\"color:#4A5BEE; font-weight:600;\"\u003e→\u003c\/span\u003e\u003cspan\u003eExact, prefix and semantic caching with thresholds\u003c\/span\u003e\n\u003c\/li\u003e\n\u003cli style=\"font-size: 13px; padding: 7px 0;  display: flex; gap: 10px;\"\u003e\n\u003cspan style=\"color:#4A5BEE; font-weight:600;\"\u003e→\u003c\/span\u003e\u003cspan\u003eA quality gate on every cost cut, with before\/after numbers\u003c\/span\u003e\n\u003c\/li\u003e\n    \u003c\/ul\u003e\n  \u003c\/div\u003e\n  \u003cdiv style=\"display:flex; align-items:center; gap:20px; background:#FFFFFF; border:1px solid #E8E6E0; border-radius:8px; padding:14px 20px; margin-bottom:24px;\"\u003e\n    \u003cspan style=\"font-size:11px; color:#888780; font-family:monospace;\"\u003e📄 devika-llm-cost-latency-optimizer.skill\u003c\/span\u003e\n    \u003cspan style=\"font-size:11px; color:#888780;\"\u003eUnder 2 min install\u003c\/span\u003e\n    \u003cspan style=\"font-size:11px; color:#888780;\"\u003eWorks with Claude, ChatGPT \u0026amp; any AI chat\u003c\/span\u003e\n  \u003c\/div\u003e\n  \u003cdiv style=\"border-left:3px solid #4A5BEE; padding-left:16px;\"\u003e\n    \u003cp style=\"font-size:10px; font-weight:600; color:#4A5BEE; letter-spacing:0.08em; text-transform:uppercase; margin:0 0 6px 0;\"\u003eHow to install\u003c\/p\u003e\n    \u003cp style=\"font-size:12px; color:#555550; line-height:1.7; margin:0;\"\u003eDownload the .skill package → open Claude → paste SKILL.md into your Project Instructions or system prompt → describe your requirement → Devika builds the answer. Includes a full worked example so you see exactly what you get.\u003c\/p\u003e\n  \u003c\/div\u003e\n\u003c\/div\u003e","brand":"KissMySkills","offers":[{"title":"Default Title","offer_id":58313022636296,"sku":null,"price":29.0,"currency_code":"USD","in_stock":true}],"thumbnail_url":"\/\/cdn.shopify.com\/s\/files\/1\/1036\/1444\/7880\/files\/ag2-devika-cover.png?v=1785164471"},{"product_id":"soren-agent-memory-knowledge-graph","title":"Soren — Agent Memory \u0026 Knowledge-Graph Engineer AI Skill","description":"\u003cdiv style=\"font-family: 'DM Sans', sans-serif; color: #1A1A18; max-width: 680px;\"\u003e\n  \u003cp style=\"font-size: 16px; font-weight: 600; line-height: 1.5; margin: 0 0 8px 0;\"\u003eDrop Soren into Claude and get a senior memory engineer who gives your agent durable, useful memory, with the write path done right before retrieval.\u003c\/p\u003e\n  \u003cp style=\"font-size: 13px; color: #555550; line-height: 1.7; margin: 0 0 28px 0;\"\u003eSoren builds agent memory: the memory types (working, episodic, semantic, procedural), the write path (fact and entity extraction, dedup, conflict resolution and supersede, decay and forgetting), the read path (retrieval by relevance and recency), storage backends (vector for semantic recall, a knowledge graph for entities and relations, key-value for profile), knowledge-graph construction and GraphRAG, personalization and multi-session continuity, privacy and deletion, and evaluating whether memory improves task success. Tools like Mem0, Zep, Letta\/MemGPT, Neo4j and pgvector, without lock-in. Bound memory growth and evaluate on a held-out set before production.\u003c\/p\u003e\n  \u003cdiv style=\"background: #ECEDFC; border-radius: 12px; padding: 24px 28px; margin-bottom: 24px;\"\u003e\n    \u003cp style=\"font-size: 10px; font-weight: 600; color: #4A5BEE; letter-spacing: 0.08em; text-transform: uppercase; margin: 0 0 16px 0;\"\u003eWhat you get\u003c\/p\u003e\n    \u003cul style=\"margin: 0; padding: 0; list-style: none;\"\u003e\n\u003cli style=\"font-size: 13px; padding: 7px 0; border-bottom: 1px solid rgba(74,91,238,0.14); display: flex; gap: 10px;\"\u003e\n\u003cspan style=\"color:#4A5BEE; font-weight:600;\"\u003e→\u003c\/span\u003e\u003cspan\u003eMemory types and the write path (extract, dedup, supersede, decay)\u003c\/span\u003e\n\u003c\/li\u003e\n\u003cli style=\"font-size: 13px; padding: 7px 0; border-bottom: 1px solid rgba(74,91,238,0.14); display: flex; gap: 10px;\"\u003e\n\u003cspan style=\"color:#4A5BEE; font-weight:600;\"\u003e→\u003c\/span\u003e\u003cspan\u003eRead path: retrieval by relevance and recency\u003c\/span\u003e\n\u003c\/li\u003e\n\u003cli style=\"font-size: 13px; padding: 7px 0; border-bottom: 1px solid rgba(74,91,238,0.14); display: flex; gap: 10px;\"\u003e\n\u003cspan style=\"color:#4A5BEE; font-weight:600;\"\u003e→\u003c\/span\u003e\u003cspan\u003eHybrid backends: vector + knowledge graph + profile\u003c\/span\u003e\n\u003c\/li\u003e\n\u003cli style=\"font-size: 13px; padding: 7px 0;  display: flex; gap: 10px;\"\u003e\n\u003cspan style=\"color:#4A5BEE; font-weight:600;\"\u003e→\u003c\/span\u003e\u003cspan\u003eGraphRAG, personalization, privacy\/deletion and memory eval\u003c\/span\u003e\n\u003c\/li\u003e\n    \u003c\/ul\u003e\n  \u003c\/div\u003e\n  \u003cdiv style=\"display:flex; align-items:center; gap:20px; background:#FFFFFF; border:1px solid #E8E6E0; border-radius:8px; padding:14px 20px; margin-bottom:24px;\"\u003e\n    \u003cspan style=\"font-size:11px; color:#888780; font-family:monospace;\"\u003e📄 soren-agent-memory-knowledge-graph.skill\u003c\/span\u003e\n    \u003cspan style=\"font-size:11px; color:#888780;\"\u003eUnder 2 min install\u003c\/span\u003e\n    \u003cspan style=\"font-size:11px; color:#888780;\"\u003eWorks with Claude, ChatGPT \u0026amp; any AI chat\u003c\/span\u003e\n  \u003c\/div\u003e\n  \u003cdiv style=\"border-left:3px solid #4A5BEE; padding-left:16px;\"\u003e\n    \u003cp style=\"font-size:10px; font-weight:600; color:#4A5BEE; letter-spacing:0.08em; text-transform:uppercase; margin:0 0 6px 0;\"\u003eHow to install\u003c\/p\u003e\n    \u003cp style=\"font-size:12px; color:#555550; line-height:1.7; margin:0;\"\u003eDownload the .skill package → open Claude → paste SKILL.md into your Project Instructions or system prompt → describe your requirement → Soren builds the answer. Includes a full worked example so you see exactly what you get.\u003c\/p\u003e\n  \u003c\/div\u003e\n\u003c\/div\u003e","brand":"KissMySkills","offers":[{"title":"Default Title","offer_id":58313026044168,"sku":null,"price":34.0,"currency_code":"USD","in_stock":true}],"thumbnail_url":"\/\/cdn.shopify.com\/s\/files\/1\/1036\/1444\/7880\/files\/ag2-soren-cover.png?v=1785164514"},{"product_id":"liora-agent-qa-red-team-engineer","title":"Liora — Agent QA \u0026 Red-Team Engineer AI Skill","description":"\u003cdiv style=\"font-family: 'DM Sans', sans-serif; color: #1A1A18; max-width: 680px;\"\u003e\n  \u003cp style=\"font-size: 16px; font-weight: 600; line-height: 1.5; margin: 0 0 8px 0;\"\u003eDrop Liora into Claude and get a senior agent QA engineer who tests and hardens your own agent with behavioral suites and a defensive red-team pass.\u003c\/p\u003e\n  \u003cp style=\"font-size: 13px; color: #555550; line-height: 1.7; margin: 0 0 28px 0;\"\u003eLiora quality-assures non-deterministic agents: behavioral test suites (assertions on tool calls, output schema, task completion and refusal behavior), an eval\/test dataset (happy path, edge cases, robustness), trajectory testing (tool selection, argument correctness, loop and termination, recovery), defensive robustness and red-team testing of your own agent (prompt-injection and jailbreak resistance, PII-leak checks, off-policy refusals), regression testing across prompt and model changes, flakiness handling (seeds, pass@k), grounding checks, and CI gates. Tools like promptfoo, DeepEval, LangSmith and pytest, without lock-in. Strictly defensive: your own agent only, no attack tooling. Gate in CI in staging before release.\u003c\/p\u003e\n  \u003cdiv style=\"background: #ECEDFC; border-radius: 12px; padding: 24px 28px; margin-bottom: 24px;\"\u003e\n    \u003cp style=\"font-size: 10px; font-weight: 600; color: #4A5BEE; letter-spacing: 0.08em; text-transform: uppercase; margin: 0 0 16px 0;\"\u003eWhat you get\u003c\/p\u003e\n    \u003cul style=\"margin: 0; padding: 0; list-style: none;\"\u003e\n\u003cli style=\"font-size: 13px; padding: 7px 0; border-bottom: 1px solid rgba(74,91,238,0.14); display: flex; gap: 10px;\"\u003e\n\u003cspan style=\"color:#4A5BEE; font-weight:600;\"\u003e→\u003c\/span\u003e\u003cspan\u003eBehavioral tests: tool calls, schema, task completion, refusals\u003c\/span\u003e\n\u003c\/li\u003e\n\u003cli style=\"font-size: 13px; padding: 7px 0; border-bottom: 1px solid rgba(74,91,238,0.14); display: flex; gap: 10px;\"\u003e\n\u003cspan style=\"color:#4A5BEE; font-weight:600;\"\u003e→\u003c\/span\u003e\u003cspan\u003eTrajectory testing and flakiness handling (pass@k)\u003c\/span\u003e\n\u003c\/li\u003e\n\u003cli style=\"font-size: 13px; padding: 7px 0; border-bottom: 1px solid rgba(74,91,238,0.14); display: flex; gap: 10px;\"\u003e\n\u003cspan style=\"color:#4A5BEE; font-weight:600;\"\u003e→\u003c\/span\u003e\u003cspan\u003eDefensive red-team: injection\/jailbreak resistance, PII-leak checks\u003c\/span\u003e\n\u003c\/li\u003e\n\u003cli style=\"font-size: 13px; padding: 7px 0;  display: flex; gap: 10px;\"\u003e\n\u003cspan style=\"color:#4A5BEE; font-weight:600;\"\u003e→\u003c\/span\u003e\u003cspan\u003eRegression testing and CI gates before release\u003c\/span\u003e\n\u003c\/li\u003e\n    \u003c\/ul\u003e\n  \u003c\/div\u003e\n  \u003cdiv style=\"display:flex; align-items:center; gap:20px; background:#FFFFFF; border:1px solid #E8E6E0; border-radius:8px; padding:14px 20px; margin-bottom:24px;\"\u003e\n    \u003cspan style=\"font-size:11px; color:#888780; font-family:monospace;\"\u003e📄 liora-agent-qa-red-team-engineer.skill\u003c\/span\u003e\n    \u003cspan style=\"font-size:11px; color:#888780;\"\u003eUnder 2 min install\u003c\/span\u003e\n    \u003cspan style=\"font-size:11px; color:#888780;\"\u003eWorks with Claude, ChatGPT \u0026amp; any AI chat\u003c\/span\u003e\n  \u003c\/div\u003e\n  \u003cdiv style=\"border-left:3px solid #4A5BEE; padding-left:16px;\"\u003e\n    \u003cp style=\"font-size:10px; font-weight:600; color:#4A5BEE; letter-spacing:0.08em; text-transform:uppercase; margin:0 0 6px 0;\"\u003eHow to install\u003c\/p\u003e\n    \u003cp style=\"font-size:12px; color:#555550; line-height:1.7; margin:0;\"\u003eDownload the .skill package → open Claude → paste SKILL.md into your Project Instructions or system prompt → describe your requirement → Liora builds the answer. Includes a full worked example so you see exactly what you get.\u003c\/p\u003e\n  \u003c\/div\u003e\n\u003c\/div\u003e","brand":"KissMySkills","offers":[{"title":"Default Title","offer_id":58313031516424,"sku":null,"price":29.0,"currency_code":"USD","in_stock":true}],"thumbnail_url":"\/\/cdn.shopify.com\/s\/files\/1\/1036\/1444\/7880\/files\/ag2-liora-cover.png?v=1785164557"},{"product_id":"cyrus-ai-agent-architect","title":"Cyrus — AI Agent Architect AI Skill","description":"\u003cdiv style=\"font-family: 'DM Sans', sans-serif; color: #1A1A18; max-width: 680px;\"\u003e\n  \u003cp style=\"font-size: 16px; font-weight: 600; line-height: 1.5; margin: 0 0 8px 0;\"\u003eDrop Cyrus into Claude and get a senior agent architect whose signature move is talking you down the ladder to the simplest thing that works, and writing the tradeoffs into an ADR.\u003c\/p\u003e\n  \u003cp style=\"font-size: 13px; color: #555550; line-height: 1.7; margin: 0 0 28px 0;\"\u003eCyrus designs whole agentic systems and makes the hard calls: when to use a single prompt vs RAG vs a fixed workflow vs an agent vs multi-agent (agents earn their cost only with genuine open-ended decisioning), reference architectures (retrieval, tools, memory, orchestration, guardrails, evals, observability), build-vs-framework, reliability for non-deterministic systems (idempotency, retries, human-in-the-loop, fallbacks, budgets, loop guards), cost and latency architecture, security and data boundaries (least-privilege tools, untrusted-input isolation, PII), evaluation and observability as first-class, failure-mode analysis, rollout, and capturing decisions as ADRs with tradeoffs. Framework-agnostic (LangGraph, CrewAI, OpenAI Agents SDK, Anthropic tool use). Nothing ships without evals, guardrails and observability.\u003c\/p\u003e\n  \u003cdiv style=\"background: #ECEDFC; border-radius: 12px; padding: 24px 28px; margin-bottom: 24px;\"\u003e\n    \u003cp style=\"font-size: 10px; font-weight: 600; color: #4A5BEE; letter-spacing: 0.08em; text-transform: uppercase; margin: 0 0 16px 0;\"\u003eWhat you get\u003c\/p\u003e\n    \u003cul style=\"margin: 0; padding: 0; list-style: none;\"\u003e\n\u003cli style=\"font-size: 13px; padding: 7px 0; border-bottom: 1px solid rgba(74,91,238,0.14); display: flex; gap: 10px;\"\u003e\n\u003cspan style=\"color:#4A5BEE; font-weight:600;\"\u003e→\u003c\/span\u003e\u003cspan\u003eThe decision ladder: prompt vs RAG vs workflow vs agent vs multi-agent\u003c\/span\u003e\n\u003c\/li\u003e\n\u003cli style=\"font-size: 13px; padding: 7px 0; border-bottom: 1px solid rgba(74,91,238,0.14); display: flex; gap: 10px;\"\u003e\n\u003cspan style=\"color:#4A5BEE; font-weight:600;\"\u003e→\u003c\/span\u003e\u003cspan\u003eReference architecture: retrieval, tools, memory, guardrails, evals\u003c\/span\u003e\n\u003c\/li\u003e\n\u003cli style=\"font-size: 13px; padding: 7px 0; border-bottom: 1px solid rgba(74,91,238,0.14); display: flex; gap: 10px;\"\u003e\n\u003cspan style=\"color:#4A5BEE; font-weight:600;\"\u003e→\u003c\/span\u003e\u003cspan\u003eReliability, cost, latency and security tradeoffs\u003c\/span\u003e\n\u003c\/li\u003e\n\u003cli style=\"font-size: 13px; padding: 7px 0;  display: flex; gap: 10px;\"\u003e\n\u003cspan style=\"color:#4A5BEE; font-weight:600;\"\u003e→\u003c\/span\u003e\u003cspan\u003eADRs, failure-mode analysis and a phased rollout\u003c\/span\u003e\n\u003c\/li\u003e\n    \u003c\/ul\u003e\n  \u003c\/div\u003e\n  \u003cdiv style=\"display:flex; align-items:center; gap:20px; background:#FFFFFF; border:1px solid #E8E6E0; border-radius:8px; padding:14px 20px; margin-bottom:24px;\"\u003e\n    \u003cspan style=\"font-size:11px; color:#888780; font-family:monospace;\"\u003e📄 cyrus-ai-agent-architect.skill\u003c\/span\u003e\n    \u003cspan style=\"font-size:11px; color:#888780;\"\u003eUnder 2 min install\u003c\/span\u003e\n    \u003cspan style=\"font-size:11px; color:#888780;\"\u003eWorks with Claude, ChatGPT \u0026amp; any AI chat\u003c\/span\u003e\n  \u003c\/div\u003e\n  \u003cdiv style=\"border-left:3px solid #4A5BEE; padding-left:16px;\"\u003e\n    \u003cp style=\"font-size:10px; font-weight:600; color:#4A5BEE; letter-spacing:0.08em; text-transform:uppercase; margin:0 0 6px 0;\"\u003eHow to install\u003c\/p\u003e\n    \u003cp style=\"font-size:12px; color:#555550; line-height:1.7; margin:0;\"\u003eDownload the .skill package → open Claude → paste SKILL.md into your Project Instructions or system prompt → describe your requirement → Cyrus builds the answer. Includes a full worked example so you see exactly what you get.\u003c\/p\u003e\n  \u003c\/div\u003e\n\u003c\/div\u003e","brand":"KissMySkills","offers":[{"title":"Default Title","offer_id":58313034858760,"sku":null,"price":39.0,"currency_code":"USD","in_stock":true}],"thumbnail_url":"\/\/cdn.shopify.com\/s\/files\/1\/1036\/1444\/7880\/files\/ag2-cyrus-cover.png?v=1785164593"}],"url":"https:\/\/kissmyskills.com\/es\/collections\/ai-agents-llm-ops-skills.oembed","provider":"KissMySkills","version":"1.0","type":"link"}