--- title: Microsoft Agent Framework Harness and Hosted Agents... description: Microsoft's Agent Framework has reached general availability as a full production runtime, not just an SDK. The Agent Harness — the core runtime wrapping AI... canonical: https://daily.dev/posts/microsoft-agent-framework-harness-and-hosted-agents-reach-general-availability-tunrw2omo twitter:card: summary_large_image twitter:site: @dailydotdev og:type: website og:site_name: daily.dev og:title: Microsoft Agent Framework Harness and Hosted Agents Reach General Availability | daily.dev og:description: Microsoft's Agent Framework has reached general availability as a full production runtime, not just an SDK. The Agent Harness — the core runtime wrapping AI... og:url: https://daily.dev/posts/microsoft-agent-framework-harness-and-hosted-agents-reach-general-availability-tunrw2omo og:image: https://api.daily.dev/og/posts/tUNrW2OMO.png og:image:alt: Microsoft Agent Framework Harness and Hosted Agents Reach General Availability og:image:width: 1200 og:image:height: 630 og:locale: en --- > ## Documentation Index > Fetch the complete documentation index at: https://daily.dev/llms.txt > Use this file to discover all available pages before exploring further. # Microsoft Agent Framework Harness and Hosted Agents Reach General Availability **[InfoQ](https://daily.dev/sources/infoq)** · 5 min read · 5 upvotes · 0 comments ## Summary Microsoft's Agent Framework has reached general availability as a full production runtime, not just an SDK. The Agent Harness — the core runtime wrapping AI models with function invocation, history persistence, context compaction, planning, tool approval, and built-in OpenTelemetry — is now GA alongside Foundry Hosted Agents. Build 2026 also brought stable connectors for GitHub Copilot SDK and Claude Agent SDK, enabling governed multi-agent orchestration without custom adapters. A key insight from an MBZUAI analysis of Claude Code's source is that ~98.4% of a real-world agent codebase is harness infrastructure, not AI logic — validating the harness-first approach. Orchestration patterns (sequential, parallel, Magentic) share one API, and all coding-agent traffic flows through unified OpenTelemetry traces and Foundry dashboards. The framework is available in .NET and Python. ## Full article daily.dev links to this article rather than hosting it. Read it at the original source: ## Questions this post answers ### What features does the Microsoft Agent Framework Harness include now that it's generally available? The Agent Harness, generally available following Build 2026, includes function invocation, per-call history persistence, context compaction, a todo list with plan and execute modes, file memory, skills, web search, tool approval, and built-in OpenTelemetry, all enabled by default and individually removable. Shell tooling, file access, background sub-agents, and automatic looping remain opt-in and emit warnings when enabled. _Track how the agent harness landscape evolves as teams shift from building agents to governing them on daily.dev._ ### How does Microsoft Agent Framework's runaway loop protection compare to the GitHub Copilot SDK? In an independent benchmark holding the model fixed, Agent Framework halted its own loop after 40 round-trips and returned a limit-reached message, while the GitHub Copilot SDK, with host-side stopping controls off, ran to 300 round-trips without stopping on its own. Both reached identical answers in the same number of steps otherwise; the difference was in runtime safety design, not reasoning. _Compare agent runtime safety trade-offs like this before picking a harness, tracked on daily.dev._ ### How much of an AI coding agent's codebase is actually the harness versus the model's decision logic? An analysis of a leaked Claude Code v2.1.88 bundle (roughly 512,000 lines across 1,884 files, briefly exposed via an npm sourcemap on March 31) estimated about 98.4% of the codebase is harness infrastructure such as permissions, context management, sandboxing, tool routing, and recovery, with only about 1.6% being AI decision logic. The authors note this is a line-count classification of a leak-derived, partly generated bundle, not a universal audit. _Follow findings like this on daily.dev to understand what really makes coding agents work._ ## Similar posts on daily.dev - [From Local to Production: Deploy Your Microsoft Agent Framework Agent with Foundry Hosted Agents](https://daily.dev/posts/from-local-to-production-deploy-your-microsoft-agent-framework-agent-with-foundry-hosted-agents-xrsxs4zxt) · DevBlogs · 1 upvotes · 0 comments - [Microsoft Foundry Now Builds, Deploys, and Governs AI Agents End to End.](https://daily.dev/posts/microsoft-foundry-now-builds-deploys-and-governs-ai-agents-end-to-end--it7nmfneb) · ITNEXT · 1 upvotes · 0 comments - [Microsoft Agent Framework \(Preview\): AI Agent Development for .NET](https://daily.dev/posts/microsoft-agent-framework-preview-ai-agent-development-for-net-uwehk5y4a) · Syncfusion · 2 upvotes · 0 comments --- Tags: [#microsoft](https://daily.dev/tags/microsoft), [#ai-agents](https://daily.dev/tags/ai-agents), [#azure](https://daily.dev/tags/azure) [View this post on daily.dev](https://daily.dev/posts/microsoft-agent-framework-harness-and-hosted-agents-reach-general-availability-tunrw2omo) ```json {"@context":"https://schema.org","@graph":[{"@type":"Organization","@id":"https://daily.dev/#organization","name":"daily.dev","url":"https://daily.dev","logo":{"@type":"ImageObject","url":"https://daily.dev/apple-touch-icon.png","width":180,"height":180},"sameAs":["https://twitter.com/dailydotdev","https://github.com/dailydotdev","https://www.linkedin.com/company/daily-dev-ltd"]},{"@type":"WebSite","@id":"https://daily.dev/#website","url":"https://daily.dev","name":"daily.dev","publisher":{"@id":"https://daily.dev/#organization"},"potentialAction":{"@type":"SearchAction","target":{"@type":"EntryPoint","urlTemplate":"https://daily.dev/search?q={search_term_string}"},"query-input":"required name=search_term_string"}}]} {"@context":"https://schema.org","@type":"TechArticle","headline":"Microsoft Agent Framework Harness and Hosted Agents Reach General Availability","url":"https://daily.dev/posts/microsoft-agent-framework-harness-and-hosted-agents-reach-general-availability-tunrw2omo","mainEntityOfPage":{"@type":"WebPage","@id":"https://daily.dev/posts/microsoft-agent-framework-harness-and-hosted-agents-reach-general-availability-tunrw2omo"},"datePublished":"2026-08-03T10:46:45.282Z","dateModified":"2026-09-14T06:15:47.410Z","description":"Microsoft's Agent Framework has reached general availability as a full production runtime, not just an SDK. The Agent Harness — the core runtime wrapping AI...","image":"https://media.daily.dev/image/upload/f_auto,q_auto/v1/posts/02a93c15fcd253ebcfe38b119f08e525?_a=AQAEuop","thumbnailUrl":"https://media.daily.dev/image/upload/f_auto,q_auto/v1/posts/02a93c15fcd253ebcfe38b119f08e525?_a=AQAEuop","isAccessibleForFree":true,"articleSection":"InfoQ","inLanguage":"en","publisher":{"@type":"Organization","name":"daily.dev","url":"https://daily.dev","logo":{"@type":"ImageObject","url":"https://daily.dev/apple-touch-icon.png","width":180,"height":180}},"author":{"@type":"Organization","name":"InfoQ","logo":"https://media.daily.dev/image/upload/t_logo,f_auto/v1/logos/afc3bced3e1e4b188dd9127017a60e0c","url":"https://daily.dev/sources/infoq"},"commentCount":0,"discussionUrl":"https://daily.dev/posts/microsoft-agent-framework-harness-and-hosted-agents-reach-general-availability-tunrw2omo","interactionStatistic":[{"@type":"InteractionCounter","interactionType":{"@type":"LikeAction"},"userInteractionCount":5},{"@type":"InteractionCounter","interactionType":{"@type":"CommentAction"},"userInteractionCount":0}],"keywords":"microsoft,ai-agents,azure","timeRequired":"PT5M"} {"@context":"https://schema.org","@type":"BreadcrumbList","itemListElement":[{"@type":"ListItem","position":1,"name":"Home","item":"https://daily.dev"},{"@type":"ListItem","position":2,"name":"InfoQ","item":"https://daily.dev/sources/infoq"},{"@type":"ListItem","position":3,"name":"Microsoft Agent Framework Harness and Hosted Agents Reach General Availability"}]} {"@context":"https://schema.org","@type":"FAQPage","@id":"https://daily.dev/posts/microsoft-agent-framework-harness-and-hosted-agents-reach-general-availability-tunrw2omo#faq","mainEntity":[{"@type":"Question","name":"What features does the Microsoft Agent Framework Harness include now that it's generally available?","acceptedAnswer":{"@type":"Answer","text":"The Agent Harness, generally available following Build 2026, includes function invocation, per-call history persistence, context compaction, a todo list with plan and execute modes, file memory, skills, web search, tool approval, and built-in OpenTelemetry, all enabled by default and individually removable. Shell tooling, file access, background sub-agents, and automatic looping remain opt-in and emit warnings when enabled. Track how the agent harness landscape evolves as teams shift from building agents to governing them on daily.dev."}},{"@type":"Question","name":"How does Microsoft Agent Framework's runaway loop protection compare to the GitHub Copilot SDK?","acceptedAnswer":{"@type":"Answer","text":"In an independent benchmark holding the model fixed, Agent Framework halted its own loop after 40 round-trips and returned a limit-reached message, while the GitHub Copilot SDK, with host-side stopping controls off, ran to 300 round-trips without stopping on its own. Both reached identical answers in the same number of steps otherwise; the difference was in runtime safety design, not reasoning. Compare agent runtime safety trade-offs like this before picking a harness, tracked on daily.dev."}},{"@type":"Question","name":"How much of an AI coding agent's codebase is actually the harness versus the model's decision logic?","acceptedAnswer":{"@type":"Answer","text":"An analysis of a leaked Claude Code v2.1.88 bundle (roughly 512,000 lines across 1,884 files, briefly exposed via an npm sourcemap on March 31) estimated about 98.4% of the codebase is harness infrastructure such as permissions, context management, sandboxing, tool routing, and recovery, with only about 1.6% being AI decision logic. The authors note this is a line-count classification of a leak-derived, partly generated bundle, not a universal audit. Follow findings like this on daily.dev to understand what really makes coding agents work."}}]} ```