[{"data":1,"prerenderedAt":-1},["ShallowReactive",2],{"skill:01M7EMN8PSFYHTYBQGP7KASHG9":3,"skill-same:01M7EMN8PSFYHTYBQGP7KASHG9":47},{"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":25,"commitSha":26,"skillDirectory":27,"forksCount":28,"repositoryHtmlUrl":29,"repositoryHomepageUrl":6,"ownerHtmlUrl":30,"ownerBlogUrl":31,"ownerDescription":6,"updatedAt":32,"install":33,"resources":46},"01M7EMN8PSFYHTYBQGP7KASHG9","context-engineering",null,"Engineer the full context a model sees — system prompts, retrieved knowledge, tool definitions, state, and history as one designed artifact. Use when moving beyond prompt tweaks to systematic context design.","TanNhatCMS\u002Fjarvis","TanNhatCMS","https:\u002F\u002Favatars.githubusercontent.com\u002Fu\u002F113796420?v=4","01M753MV17KG5CD3152G3WQEQY","KietNT","01M7EKEDYVYQ6PWVJJMZ911VMX",0,"MIT","https:\u002F\u002Fgithub.com\u002FTanNhatCMS\u002Fjarvis\u002Fblob\u002Fmain\u002Fskills\u002Fams-context-engineering\u002FSKILL.md",[],[],"2026-11-15 14:19:17",false,"---\nname: context-engineering\ndescription: Engineer the full context a model sees — system prompts, retrieved knowledge, tool definitions, state, and history as one designed artifact. Use when moving beyond prompt tweaks to systematic context design.\ncategory: ai-research\n---\n\n# Context Engineering\n\nPrompt engineering optimizes the words; context engineering designs everything the model sees — \ninstructions, knowledge, tools, memory, state, and history — as one coherent, budgeted artifact. \nIt's the discipline behind reliable agents.\n\n## Overview\n\nEvery model call receives a context: system instructions, user input, retrieved documents, tool \nschemas, conversation history, working state. Each component competes for attention and tokens. \nContext engineering treats this as a design problem: what goes in, in what order, in what form, \nwith what budget — and how it evolves across turns. The output is a context architecture: \nspecified, measured, and iterated like any system design.\n\n## When to use\n\n- Agent reliability plateaus despite prompt tuning — the problem is usually context, not wording.\n- Designing a new agent or assistant from scratch: specify the context before the prompts.\n- Debugging \"the model forgot \u002F ignored \u002F got distracted\" issues.\n- Optimizing cost and latency: context is where the tokens (and money) go.\n\n## Core concepts\n\n- **Context components**: instructions (durable behavior), knowledge (retrieved facts), tools \n(available actions), state (current task progress), history (what happened). Name each; budget each.\n- **Assembly pipeline**: how context gets built per turn — retrieve, rank, truncate, order, \ninject. A real pipeline with stages, not string concatenation.\n- **Ordering**: instructions first, task-relevant state next, evidence in the middle, immediate \nquery last. Order encodes priority for attention.\n- **Dynamic context**: context that changes per turn — retrieved passages for this step, tools \nrelevant to this phase, summaries replacing old history. Static contexts rot.\n- **Budgets**: per-component token budgets enforced in code. When retrieval overflows, truncate by \nrelevance — don't let one component starve the rest.\n- **Context observability**: log the assembled context (or its digest) per call. When the model \nmisbehaves, inspect what it actually saw.\n\n## Practical workflow\n\n1. Inventory the context: list every component currently fed to the model and its typical token \ncost.\n2. Set budgets per component based on measured value — cut or compress the lowest-value tokens \nfirst.\n3. Design the assembly pipeline: retrieval → ranking → budget enforcement → ordering → \ninjection.\n4. Make it dynamic: per-turn retrieval, phase-appropriate tools, summarized history, maintained \nstate.\n5. Add observability: log component sizes and a digest of content per call; sample full contexts \nfor review.\n6. Iterate against evals: change one component at a time, measure quality and cost deltas.\n\n```text\nContext architecture spec:\nINSTRUCTIONS:  2k tokens — role, guardrails (static, front-loaded)\nSTATE:         3k — {goal, plan, progress} (rewritten per turn)\nTOOLS:         4k — phase-relevant subset (dynamic)\nKNOWLEDGE:    15k — retrieved per-step, relevance-ranked (dynamic)\nHISTORY:      8k — recency window + rolling summary\nBUDGET:       32k soft cap — truncate knowledge first, never instructions\n```\n\n## Common pitfalls\n\n- **Prompt-only thinking**: tuning words while the context architecture is broken. Zoom out to the \nfull assembly.\n- **Static contexts**: the same giant blob every turn. Retrieve per-step; summarize history; phase \nyour tools.\n- **Unbudgeted growth**: components growing until quality collapses. Enforce budgets in code, not \nin hope.\n- **No observability**: debugging model behavior without seeing its context. Log what it saw.\n- **Attention naivety**: assuming all context is attended equally. Order deliberately; test \nplacement effects.\n- **Over-engineering early**: a full dynamic pipeline for a simple task. Start with a spec and \nbudgets; add machinery when evals justify it.\n","# Context Engineering\n\nPrompt engineering optimizes the words; context engineering designs everything the model sees — \ninstructions, knowledge, tools, memory, state, and history — as one coherent, budgeted artifact. \nIt's the discipline behind reliable agents.\n\n## Overview\n\nEvery model call receives a context: system instructions, user input, retrieved documents, tool \nschemas, conversation history, working state. Each component competes for attention and tokens. \nContext engineering treats this as a design problem: what goes in, in what order, in what form, \nwith what budget — and how it evolves across turns. The output is a context architecture: \nspecified, measured, and iterated like any system design.\n\n## When to use\n\n- Agent reliability plateaus despite prompt tuning — the problem is usually context, not wording.\n- Designing a new agent or assistant from scratch: specify the context before the prompts.\n- Debugging \"the model forgot \u002F ignored \u002F got distracted\" issues.\n- Optimizing cost and latency: context is where the tokens (and money) go.\n\n## Core concepts\n\n- **Context components**: instructions (durable behavior), knowledge (retrieved facts), tools \n(available actions), state (current task progress), history (what happened). Name each; budget each.\n- **Assembly pipeline**: how context gets built per turn — retrieve, rank, truncate, order, \ninject. A real pipeline with stages, not string concatenation.\n- **Ordering**: instructions first, task-relevant state next, evidence in the middle, immediate \nquery last. Order encodes priority for attention.\n- **Dynamic context**: context that changes per turn — retrieved passages for this step, tools \nrelevant to this phase, summaries replacing old history. Static contexts rot.\n- **Budgets**: per-component token budgets enforced in code. When retrieval overflows, truncate by \nrelevance — don't let one component starve the rest.\n- **Context observability**: log the assembled context (or its digest) per call. When the model \nmisbehaves, inspect what it actually saw.\n\n## Practical workflow\n\n1. Inventory the context: list every component currently fed to the model and its typical token \ncost.\n2. Set budgets per component based on measured value — cut or compress the lowest-value tokens \nfirst.\n3. Design the assembly pipeline: retrieval → ranking → budget enforcement → ordering → \ninjection.\n4. Make it dynamic: per-turn retrieval, phase-appropriate tools, summarized history, maintained \nstate.\n5. Add observability: log component sizes and a digest of content per call; sample full contexts \nfor review.\n6. Iterate against evals: change one component at a time, measure quality and cost deltas.\n\n```text\nContext architecture spec:\nINSTRUCTIONS:  2k tokens — role, guardrails (static, front-loaded)\nSTATE:         3k — {goal, plan, progress} (rewritten per turn)\nTOOLS:         4k — phase-relevant subset (dynamic)\nKNOWLEDGE:    15k — retrieved per-step, relevance-ranked (dynamic)\nHISTORY:      8k — recency window + rolling summary\nBUDGET:       32k soft cap — truncate knowledge first, never instructions\n```\n\n## Common pitfalls\n\n- **Prompt-only thinking**: tuning words while the context architecture is broken. Zoom out to the \nfull assembly.\n- **Static contexts**: the same giant blob every turn. Retrieve per-step; summarize history; phase \nyour tools.\n- **Unbudgeted growth**: components growing until quality collapses. Enforce budgets in code, not \nin hope.\n- **No observability**: debugging model behavior without seeing its context. Log what it saw.\n- **Attention naivety**: assuming all context is attended equally. Order deliberately; test \nplacement effects.\n- **Over-engineering early**: a full dynamic pipeline for a simple task. Start with a spec and \nbudgets; add machinery when evals justify it.\n",{"name":5,"category":24,"description":7},"ai-research","main","af388a77c2fe5e0b4fd879cca4300a0ffa9a6579","skills\u002Fams-context-engineering",1,"https:\u002F\u002Fgithub.com\u002FTanNhatCMS\u002Fjarvis","https:\u002F\u002Fgithub.com\u002FTanNhatCMS","https:\u002F\u002Fmrkiet.name.vn","2026-11-07 14:23:08",{"source":8,"skillName":5,"commands":34,"prompt":44,"zipAvailable":45},[35,38,41],{"tool":36,"command":37},"npx","npx skills add TanNhatCMS\u002Fjarvis --skill context-engineering",{"tool":39,"command":40},"bunx","bunx skills add TanNhatCMS\u002Fjarvis --skill context-engineering",{"tool":42,"command":43},"pnpm","pnpm dlx skills add TanNhatCMS\u002Fjarvis --skill context-engineering","安装技能，技能地址：https:\u002F\u002Fgithub.com\u002FTanNhatCMS\u002Fjarvis\u002Fblob\u002Fmain\u002Fskills\u002Fams-context-engineering\u002FSKILL.md",true,[],[48,55,63,70,77,84],{"skillNo":49,"name":50,"nameZh":6,"description":51,"descriptionZh":6,"repositoryFullName":8,"ownerLogin":9,"ownerAvatarUrl":10,"ownerAccountNo":11,"ownerDisplayName":12,"repositoryNo":13,"starsCount":14,"licenseSpdxId":15,"licenseFileUrl":6,"sourceHtmlUrl":52,"categories":53,"keywords":54,"publishedAt":19},"01M7EWYJR0J8DARY7E59Z6Y9MV","web-scrape","Turn web pages into LLM-ready markdown via the Firecrawl API (renders JS, anti-bot handling). Use when browser_open\u002Ffetch fails on JS-heavy pages, or when you need clean markdown\u002Fstructured extraction from a site. Requires FIRECRAWL_API_KEY.","https:\u002F\u002Fgithub.com\u002FTanNhatCMS\u002Fjarvis\u002Fblob\u002Fmain\u002Fskills\u002Fweb-scrape\u002FSKILL.md",[],[],{"skillNo":56,"name":57,"nameZh":6,"description":58,"descriptionZh":6,"repositoryFullName":8,"ownerLogin":9,"ownerAvatarUrl":10,"ownerAccountNo":11,"ownerDisplayName":12,"repositoryNo":13,"starsCount":14,"licenseSpdxId":15,"licenseFileUrl":59,"sourceHtmlUrl":60,"categories":61,"keywords":62,"publishedAt":19},"01M7EWYBN5P6F2AJG3FR24T9BP","design-taste-frontend","tasteskill v2, the anti-slop frontend skill. Art direction for websites, distilled from a study of 247 curated gallery sites (One Page Love, Awwwards Sites of the Day). Deals researched directions (idea, composition, type, palette, photographs, section blueprints) from a library, enforces a strict no-slop house style with a real-browser inspector, and runs a look-and-fix loop on screenshots. Use this whenever you build, design, redesign, restyle, polish or review a landing page, marketing site, portfolio, product page, one-pager, hero or any public-facing web page, in any stack (HTML\u002FCSS, React, Next.js, Vue, Svelte, Astro, Tailwind) — also when the user says a page looks generic, templated, \"AI-made\", sloppy, or asks to make it look designed, premium, award-level or less like every other site. Not for dashboards or app screens.","https:\u002F\u002Fgithub.com\u002FTanNhatCMS\u002Fjarvis\u002Fblob\u002Fmain\u002Fskills\u002Ftaste-design\u002FLICENSE","https:\u002F\u002Fgithub.com\u002FTanNhatCMS\u002Fjarvis\u002Fblob\u002Fmain\u002Fskills\u002Ftaste-design\u002FSKILL.md",[],[],{"skillNo":64,"name":65,"nameZh":6,"description":66,"descriptionZh":6,"repositoryFullName":8,"ownerLogin":9,"ownerAvatarUrl":10,"ownerAccountNo":11,"ownerDisplayName":12,"repositoryNo":13,"starsCount":14,"licenseSpdxId":15,"licenseFileUrl":6,"sourceHtmlUrl":67,"categories":68,"keywords":69,"publishedAt":19},"01M7EWY8FQFNQ2GB7NZZWGSSDX","test-driven-development","Use when implementing any feature or bugfix, before writing implementation code","https:\u002F\u002Fgithub.com\u002FTanNhatCMS\u002Fjarvis\u002Fblob\u002Fmain\u002Fskills\u002Fsuperpowers-tdd\u002FSKILL.md",[],[],{"skillNo":71,"name":72,"nameZh":6,"description":73,"descriptionZh":6,"repositoryFullName":8,"ownerLogin":9,"ownerAvatarUrl":10,"ownerAccountNo":11,"ownerDisplayName":12,"repositoryNo":13,"starsCount":14,"licenseSpdxId":15,"licenseFileUrl":6,"sourceHtmlUrl":74,"categories":75,"keywords":76,"publishedAt":19},"01M7EWY5SHBFRH1HDHVWM9K7TV","systematic-debugging","Use when encountering any bug, test failure, or unexpected behavior, before proposing fixes","https:\u002F\u002Fgithub.com\u002FTanNhatCMS\u002Fjarvis\u002Fblob\u002Fmain\u002Fskills\u002Fsuperpowers-debugging\u002FSKILL.md",[],[],{"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":83,"publishedAt":19},"01M7EWY3NM0Z0JAKS7HDZ24K66","requesting-code-review","Use when completing tasks, implementing major features, or before merging to verify work meets requirements","https:\u002F\u002Fgithub.com\u002FTanNhatCMS\u002Fjarvis\u002Fblob\u002Fmain\u002Fskills\u002Fsuperpowers-code-review\u002FSKILL.md",[],[],{"skillNo":85,"name":86,"nameZh":6,"description":87,"descriptionZh":6,"repositoryFullName":8,"ownerLogin":9,"ownerAvatarUrl":10,"ownerAccountNo":11,"ownerDisplayName":12,"repositoryNo":13,"starsCount":14,"licenseSpdxId":15,"licenseFileUrl":6,"sourceHtmlUrl":88,"categories":89,"keywords":90,"publishedAt":19},"01M7EWY0WRMWX4M9ZFY0WKC593","brainstorming","You MUST use this before any creative work - creating features, building components, adding functionality, modifying behavior, or planning anything new, in software or out of it (a talk, a business, a renovation).","https:\u002F\u002Fgithub.com\u002FTanNhatCMS\u002Fjarvis\u002Fblob\u002Fmain\u002Fskills\u002Fsuperpowers-brainstorming\u002FSKILL.md",[],[]]