[{"data":1,"prerenderedAt":-1},["ShallowReactive",2],{"skill:01M6X00VV8GSAJZ4MASK7T8CQF":3,"skill-same:01M6X00VV8GSAJZ4MASK7T8CQF":71},{"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":23,"publishedAt":43,"linkOnly":44,"markdownContent":45,"markdownBody":46,"frontMatter":47,"refName":48,"commitSha":49,"skillDirectory":50,"forksCount":51,"repositoryHtmlUrl":52,"repositoryHomepageUrl":6,"ownerHtmlUrl":53,"ownerBlogUrl":54,"ownerDescription":55,"updatedAt":56,"install":57,"resources":70},"01M6X00VV8GSAJZ4MASK7T8CQF","agent-evaluation",null,"Help a learner evaluate and observe an AI agent on Microsoft Foundry — quality\u002Fsafety evaluators, tracing, and reading results. Use for Lesson 3 tasks: measuring agent performance, adding observability, or interpreting evaluation output.","microsoft\u002FBuilding-AI-Agents-From-Zero-To-Production","microsoft","https:\u002F\u002Favatars.githubusercontent.com\u002Fu\u002F6154722?v=4","01M698VYDXZ6JW5WH3CRV650R8","Microsoft","01M6X00M2R68PC89FEP5PT07MG",198,"MIT","https:\u002F\u002Fgithub.com\u002Fmicrosoft\u002FBuilding-AI-Agents-From-Zero-To-Production\u002Fblob\u002Fmain\u002F.github\u002Fskills\u002Fagent-evaluation\u002FSKILL.md",[18],{"categoryNo":19,"categoryName":20,"taxonomyType":21,"taxonomyTypeDesc":22},"01JBSEEDCAT000000000000023","Skill与智能体开发","SCENARIO","使用场景",[24,25,26,27,28,29,30,31,32,33,34,35,36,37,38,39,40,41,9,42],"agent evaluation","Evaluation","quality","evaluate","for","reading","output","learner","lesson","observe","use","results","performance","Safety","observability","help","tasks","foundry","ai agent","2026-11-10 10:12:30",false,"---\nname: agent-evaluation\ndescription: >-\n  Help a learner evaluate and observe an AI agent on Microsoft Foundry — quality\u002Fsafety evaluators,\n  tracing, and reading results. Use for Lesson 3 tasks: measuring agent performance, adding\n  observability, or interpreting evaluation output.\n---\n\n# Agent evaluation & observability\n\nGuidance for measuring and improving agents (Lesson 3).\n\n## Guardrails (always apply)\n- Platform: Microsoft Foundry; framework: Microsoft Agent Framework.\n- Use the **evaluation\u002Fobservability APIs from the current `agent-framework` surface** (Foundry\n  evals + tracing). Model for any judge\u002Feval step: **`gpt-5.1`** (never `gpt-4o` \u002F `gpt-4.1`).\n- Endpoints\u002Fmodels from `.env`; `az login` for auth.\n\n## Concepts a learner should grasp\n1. **Why evaluate:** agents are non-deterministic — you need repeatable metrics (groundedness,\n   relevance, coherence, safety) to know whether a change helped or hurt.\n2. **Observability vs evaluation:** tracing captures *what happened* on each run (spans, tool calls,\n   tokens); evaluation *scores* those runs against criteria.\n3. **Close the loop:** trace → evaluate → adjust instructions\u002Ftools → re-evaluate.\n\n## How to help\n- Point the learner at `lesson-3-agent-evals\u002F` and its README for the exact evaluator setup.\n- Encourage a small, fixed test set of prompts so results are comparable across runs.\n- When an eval fails, look at the trace first (bad tool call? missing context? weak instructions?)\n  before changing the model.\n- Keep evaluation runs cheap: small datasets, and delete resources afterward (see cost & cleanup).\n\n## Validate\n- `python -m py_compile \u003Cfile>.py`.\n- An evaluation run should produce per-criterion scores you can compare between versions.\n\n## References\n- `lesson-3-agent-evals\u002FREADME.md`\n- `MIGRATION-GUIDE.md` (current SDK patterns)\n","# Agent evaluation & observability\n\nGuidance for measuring and improving agents (Lesson 3).\n\n## Guardrails (always apply)\n- Platform: Microsoft Foundry; framework: Microsoft Agent Framework.\n- Use the **evaluation\u002Fobservability APIs from the current `agent-framework` surface** (Foundry\n  evals + tracing). Model for any judge\u002Feval step: **`gpt-5.1`** (never `gpt-4o` \u002F `gpt-4.1`).\n- Endpoints\u002Fmodels from `.env`; `az login` for auth.\n\n## Concepts a learner should grasp\n1. **Why evaluate:** agents are non-deterministic — you need repeatable metrics (groundedness,\n   relevance, coherence, safety) to know whether a change helped or hurt.\n2. **Observability vs evaluation:** tracing captures *what happened* on each run (spans, tool calls,\n   tokens); evaluation *scores* those runs against criteria.\n3. **Close the loop:** trace → evaluate → adjust instructions\u002Ftools → re-evaluate.\n\n## How to help\n- Point the learner at `lesson-3-agent-evals\u002F` and its README for the exact evaluator setup.\n- Encourage a small, fixed test set of prompts so results are comparable across runs.\n- When an eval fails, look at the trace first (bad tool call? missing context? weak instructions?)\n  before changing the model.\n- Keep evaluation runs cheap: small datasets, and delete resources afterward (see cost & cleanup).\n\n## Validate\n- `python -m py_compile \u003Cfile>.py`.\n- An evaluation run should produce per-criterion scores you can compare between versions.\n\n## References\n- `lesson-3-agent-evals\u002FREADME.md`\n- `MIGRATION-GUIDE.md` (current SDK patterns)\n",{"name":5,"description":7},"main","8c4367c3df7138fa869c386fda70f9de89f23c75",".github\u002Fskills\u002Fagent-evaluation",72,"https:\u002F\u002Fgithub.com\u002Fmicrosoft\u002FBuilding-AI-Agents-From-Zero-To-Production","https:\u002F\u002Fgithub.com\u002Fmicrosoft","https:\u002F\u002Fopensource.microsoft.com","Open source projects and samples from Microsoft","2026-10-31 01:14:49",{"source":8,"skillName":5,"commands":58,"prompt":68,"zipAvailable":69},[59,62,65],{"tool":60,"command":61},"npx","npx skills add microsoft\u002FBuilding-AI-Agents-From-Zero-To-Production --skill agent-evaluation",{"tool":63,"command":64},"bunx","bunx skills add microsoft\u002FBuilding-AI-Agents-From-Zero-To-Production --skill agent-evaluation",{"tool":66,"command":67},"pnpm","pnpm dlx skills add microsoft\u002FBuilding-AI-Agents-From-Zero-To-Production --skill agent-evaluation","安装技能，技能地址：https:\u002F\u002Fgithub.com\u002Fmicrosoft\u002FBuilding-AI-Agents-From-Zero-To-Production\u002Fblob\u002Fmain\u002F.github\u002Fskills\u002Fagent-evaluation\u002FSKILL.md",true,[],[72,95,116,136,156],{"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":6,"sourceHtmlUrl":76,"categories":77,"keywords":79,"publishedAt":43},"01M6X01NFNWB6R9QV6137VGMS0","multi-agent-a2a","Help a learner compose agents as networked services using the open Agent-to-Agent (A2A) protocol — expose an agent as an A2A server and consume a remote agent as a peer. Use for Lesson 7 tasks: multi-agent orchestration and A2A client\u002Fserver wiring.","https:\u002F\u002Fgithub.com\u002Fmicrosoft\u002FBuilding-AI-Agents-From-Zero-To-Production\u002Fblob\u002Fmain\u002F.github\u002Fskills\u002Fmulti-agent-a2a\u002FSKILL.md",[78],{"categoryNo":19,"categoryName":20,"taxonomyType":21,"taxonomyTypeDesc":22},[80,81,82,83,84,85,28,86,31,87,88,32,89,90,34,91,92,93,40,94],"multi agent a2a","multi","a2a","multi agent","agent orchestration","wiring","remote","as","agent-to-agent","peer","compose","open","services","server","orchestration",{"skillNo":96,"name":97,"nameZh":6,"description":98,"descriptionZh":6,"repositoryFullName":8,"ownerLogin":9,"ownerAvatarUrl":10,"ownerAccountNo":11,"ownerDisplayName":12,"repositoryNo":13,"starsCount":14,"licenseSpdxId":15,"licenseFileUrl":6,"sourceHtmlUrl":99,"categories":100,"keywords":102,"publishedAt":43},"01M6X01FJ3ZV88JAAF73R9Q2NQ","microsoft-toolbox","Help a learner use Microsoft Toolbox to define tools once and govern them centrally, then consume a toolbox from an agent through a single MCP endpoint. Use for Lesson 6 tasks: tool discovery, registration, governance, and versioning.","https:\u002F\u002Fgithub.com\u002Fmicrosoft\u002FBuilding-AI-Agents-From-Zero-To-Production\u002Fblob\u002Fmain\u002F.github\u002Fskills\u002Fmicrosoft-toolbox\u002FSKILL.md",[101],{"categoryNo":19,"categoryName":20,"taxonomyType":21,"taxonomyTypeDesc":22},[103,104,9,105,106,28,107,31,108,32,109,34,110,111,112,113,39,40,114,115],"microsoft toolbox","toolbox","discovery","single","tool discovery","define","registration","versioning","tool","Tools","endpoint","governance","MCP",{"skillNo":117,"name":118,"nameZh":6,"description":119,"descriptionZh":6,"repositoryFullName":8,"ownerLogin":9,"ownerAvatarUrl":10,"ownerAccountNo":11,"ownerDisplayName":12,"repositoryNo":13,"starsCount":14,"licenseSpdxId":15,"licenseFileUrl":6,"sourceHtmlUrl":120,"categories":121,"keywords":123,"publishedAt":43},"01M6X01A1BEKTXZSPN6817VJV8","mcp-integration","Help a learner give an agent tools through the Model Context Protocol (MCP) — both client-side MCP and Foundry Hosted MCP tools. Use when connecting an agent to an MCP server (e.g. the GitHub MCP server), adding a remote tool, or reasoning about MCP approval\u002Fgovernance.","https:\u002F\u002Fgithub.com\u002Fmicrosoft\u002FBuilding-AI-Agents-From-Zero-To-Production\u002Fblob\u002Fmain\u002F.github\u002Fskills\u002Fmcp-integration\u002FSKILL.md",[122],{"categoryNo":19,"categoryName":20,"taxonomyType":21,"taxonomyTypeDesc":22},[124,125,115,126,127,86,39,31,128,34,129,111,130,93,131,132,114,133,134,135],"MCP Integration","integration","when","about","github mcp","model","github mcp server","protocol","reasoning","model context protocol","mcp server","github",{"skillNo":137,"name":138,"nameZh":6,"description":139,"descriptionZh":6,"repositoryFullName":8,"ownerLogin":9,"ownerAvatarUrl":10,"ownerAccountNo":11,"ownerDisplayName":12,"repositoryNo":13,"starsCount":14,"licenseSpdxId":15,"licenseFileUrl":6,"sourceHtmlUrl":140,"categories":141,"keywords":143,"publishedAt":43},"01M6X0146GACX2QRFA1TG1T0A1","hosted-agent-deployment","Help a learner deploy an agent to production as a Microsoft Foundry Hosted Agent, and decide when Capability Hosts are needed. Use for Lessons 4–5: hosted deployment, thread\u002Fmemory persistence, bring-your-own storage, and enterprise governance.","https:\u002F\u002Fgithub.com\u002Fmicrosoft\u002FBuilding-AI-Agents-From-Zero-To-Production\u002Fblob\u002Fmain\u002F.github\u002Fskills\u002Fhosted-agent-deployment\u002FSKILL.md",[142],{"categoryNo":19,"categoryName":20,"taxonomyType":21,"taxonomyTypeDesc":22},[144,145,114,146,147,31,87,34,148,149,150,39,28,151,9,152,153,126,154,155],"hosted agent deployment","deployment","hosts","bring","decide","Production","Enterprise","deploy","storage","memory","lessons","thread",{"skillNo":157,"name":158,"nameZh":6,"description":159,"descriptionZh":6,"repositoryFullName":8,"ownerLogin":9,"ownerAvatarUrl":10,"ownerAccountNo":11,"ownerDisplayName":12,"repositoryNo":13,"starsCount":14,"licenseSpdxId":15,"licenseFileUrl":6,"sourceHtmlUrl":160,"categories":161,"keywords":163,"publishedAt":43},"01M6X010DC4ZF2TAJ2NSFAPXZ0","foundry-agent-basics","Help a learner build their first AI agent with the Microsoft Agent Framework (MAF) on Microsoft Foundry. Use when the task is creating a basic agent, wiring up FoundryChatClient, choosing a model, running the DevUI, or understanding lesson 1–2 concepts (instructions, tools, threads).","https:\u002F\u002Fgithub.com\u002Fmicrosoft\u002FBuilding-AI-Agents-From-Zero-To-Production\u002Fblob\u002Fmain\u002F.github\u002Fskills\u002Ffoundry-agent-basics\u002FSKILL.md",[162],{"categoryNo":19,"categoryName":20,"taxonomyType":21,"taxonomyTypeDesc":22},[41,164,165,34,166,42,167,168,169,129,170,31,171,126,85,172,173,174,32,175],"foundry agent basics","basics","up","agent framework","microsoft agent framework","first","maf","running","threads","task","with","basic"]