# agent harness Published articles for agent harness. This is one page of public article previews, not the complete archive. Follow Next page to continue. Summaries are not the original full articles. ## Migrating the GitHub Copilot runtime to Rust, using Copilot DevFeed: [Migrating the GitHub Copilot runtime to Rust, using Copilot]() Original publisher: [Read original article]() Author: Stephen Toub Published: 2026-09-17T00:26:43Z Content type: article Language: en Sources: [GitHub Engineering]() Topics: [GitHub Copilot](), [Rust](), [Agent Harness](), [GitHub Copilot SDK](), [GitHub]() Tags: [agent-harness](), [ai-agents](), [ai-ml](), [developer-experience](), [generative-ai](), [github](), [github-copilot](), [github-copilot-sdk](), [rust]() ### AI overview The GitHub Copilot agent runtime was rewritten from TypeScript on Node.js and V8 into more than 800,000 lines of production Rust. AI agents wrote most of the code across 128 pull requests, and the runtime was shipped incrementally while regressions were fixed and performance improved by orders of magnitude. ### Source excerpt A rewrite this size wasn't affordable before agents. Here's what porting the Copilot agent runtime to 800,000 lines of production Rust actually took. The post Migrating the GitHub Copilot runtime to Rust, using Copilot appeared first on The GitHub Blog. ## Build Your Own AI Agent Harness in C#, the MafClaw Live Series DevFeed: [Build Your Own AI Agent Harness in C#, the MafClaw Live Series]() Original publisher: [Read original article]() Author: Bruno Capuano Published: 2026-09-16T21:00:00Z Content type: tutorial Language: en Sources: [.NET Blog]() Topics: [Agent Harness](), [Microsoft Agent Framework](), [AI Agent](), [C#](), [.NET]() Tags: [agent](), [agent-framework](), [agent-harness](), [ai](), [building](), [c-sharp](), [cloud](), [csharp](), [microsoft-agent-framework](), [microsoft-reactor](), [net](), [observability]() ### AI overview This article introduces a four-part live series that builds a complete C# AI agent using the Microsoft Agent Framework harness. It describes adding tools, planning, file access, approvals, memory, skills, shell commands, code execution, background agents, observability, governance, evaluations, and hosted deployment. ### Source excerpt I am building a complete C# agent live, from a single call around an IChatClient to a production-ready, observable, governed agent, using the Microsoft Agent Framework harness in a 4-part Microsoft Reactor series. The post Build Your Own AI Agent Harness in C#, the MafClaw Live Series appeared first on .NET Blog. ## Bolt.new tests Forge, offering more coding-model usage in exchange for anonymized developer sessions DevFeed: [Bolt.new tests Forge, offering more coding-model usage in exchange for anonymized developer sessions]() Original publisher: [Read original article]() Author: Amanda Caswell Published: 2026-09-15T18:47:23Z Content type: article Language: en Sources: [The New Stack]() Topics: [AI-assisted coding](), [Training AI Models](), [Open Source Models & Datasets](), [Agent Harness]() Tags: [agent](), [agent-harness](), [ai](), [ai-development](), [ai-engineering](), [ai-models](), [coding-agents](), [data](), [developer-tools](), [developers](), [training]() ### AI overview Bolt.new is testing Forge, a research preview for individual Pro subscribers that offers up to 50 times more usage of open-weight coding models in exchange for opting in to share anonymized coding sessions. The sessions may include prompts, source code, fix traces, and conversations with the coding agent, and will support an Arcee AI project to train a trillion-parameter-class open-weight model. ### Source excerpt Bolt.new, StackBlitz's browser-based AI development platform, is testing a new trade with developers: more coding-model usage in exchange for training The post Bolt is giving developers 50x more compute. But there's a catch. appeared first on The New Stack. ## Beyond the model: Engineering AI infra with scientific judgement DevFeed: [Beyond the model: Engineering AI infra with scientific judgement]() Original publisher: [Read original article]() Author: AirbnbEng Published: 2026-09-15T17:06:18Z Content type: article Language: en Sources: [The Airbnb Tech Blog - Medium]() Topics: [Agent Harness](), [AI Agent](), [Data Science](), [data](), [datasets](), [Large Language Model]() Tags: [agent-harness](), [ai](), [ai-agent](), [data](), [data-science](), [datasets](), [engineering](), [llms](), [quality](), [science](), [trust]() ### AI overview Airbnb describes an agent harness for data science that embeds scientific methodology around an AI model. The system guides agents through framing questions, selecting evidence, and recording decisions so unstructured-data investigations can be reproduced, audited, challenged, and extended across languages, geographies, and LLM-based products. ### Source excerpt How Airbnb's agent harness transforms unstructured data exploration by encoding scientific methodology into scalable, reproducible, and audit-ready infrastructure. By: Wren Dougherty Ask a coding agent to analyze 100,000 customer support conversations and within minutes you'll have a polished taxonomy, precise prevalence numbers, and an executive-ready summary. What you can't see is the investigation that produced them: the methods it chose, the evidence it weighed, how much to trust it, or whether a second request would agree. All that reaches you is the polish. The model is undeniably intelligent, but intelligence without methodology is not science. LLMs certainly make for confident scientists, but we need them to be responsible ones. Smarter models help, but intelligence has never been the whole of science, in people or in machines. The method is as much the product as the answer. That is the idea behind the agent harness we built for data science: the methodology itself, built as infrastructure around the model. It governs how an AI agent operates, from framing a question to selecting evidence to recording decisions, so results can be reproduced, audited, and challenged, and the method shared, inspected, and built on. The challenge of unstructured data exploration In 2025, Airbnb was preparing to launch an AI customer service assistant. Before it could ship, we needed to understand exactly what kinds of situations it would face in the real world. That included rare events that could be risky for AI to interact with, and involved examining their taxonomy and prevalence to create the datasets that would help us build a more responsible product. The investigative work to do this was rigorous, but the process was deeply artisanal. Months of high-touch iteration went into each investigation, from finding the right data, reviewing samples with experts, and generating representative datasets, and the method was manually curated across notebooks, tables, docs, and indiv ## AI safety does not stop at the model DevFeed: [AI safety does not stop at the model]() Original publisher: [Read original article]() Author: Samar Abbas Published: 2026-09-15T00:00:00Z Content type: opinion Language: en Sources: [Temporal Blog]() Topics: [ai safety](), [Agent Harness](), [reliability](), [retry](), [Security]() Tags: [agent](), [agent-harness](), [ai](), [reliability](), [retry](), [safety](), [security](), [temporal-voices]() ### AI overview The article argues that AI safety extends beyond model behavior to the application layer, where companies must control agent authority and enforce approvals, policies, and credential limits. It emphasizes that these controls must remain reliable through crashes, timeouts, and retries, and presents Temporal Agent Harness as an execution-layer control point around an agent SDK. ### Source excerpt Auditing what an agent did is only half the job. Companies also have to control what agents may do, and make those limits hold when systems fail. ## Your Agent Harness Needs Runtime Security DevFeed: [Your Agent Harness Needs Runtime Security]() Original publisher: [Read original article]() Author: Avi Chawla Published: 2026-09-09T20:59:58Z Content type: tutorial Language: en Sources: [Daily Dose of Data Science]() Topics: [agent observability](), [Agent Harness](), [Security](), [Open Source](), [telemetry]() Tags: [agent](), [agent-harness](), [ai-agents](), [anthropic](), [claude](), [cybersecurity](), [github](), [guide](), [local](), [monitoring](), [open-source](), [security](), [telemetry]() ### AI overview This guide presents Agent Beacon, an open-source telemetry layer for AI agents. It records tool calls, shell commands, file changes, and approval decisions as structured runtime events across supported agent harnesses, providing a live record of agent behavior for security monitoring and investigation. ### Source excerpt A 100% local, open-source guide to recording what AI agents actually do at runtime. ## Why Short AI Coding Prompts Can Cost You More Time DevFeed: [Why Short AI Coding Prompts Can Cost You More Time]() Original publisher: [Read original article]() Author: Dean Hume Published: 2026-09-03T09:17:56Z Content type: opinion Language: en Sources: [Dean Hume]() Topics: [Prompt Engineering](), [ai-coding](), [GitHub Copilot](), [Artificial Intelligence](), [Agent Harness](), [GitHub Copilot CLI](), [Testing]() Tags: [agent-harness](), [ai](), [ai-coding](), [coding](), [copilot](), [github-copilot](), [prompt](), [testing]() ### AI overview This opinion article connects GitHub Copilot team's findings about compressing tool output with everyday AI coding prompts. It argues that optimizing a visible step, such as shortening a response or prompt, can increase total time and cost when missing details force follow-up questions, reruns, or recovery work. It recommends giving coding assistants the full problem and relevant source material, and treating instruction files carefully because rewriting guidance can change agent behavior. ### Source excerpt Ever tried to save time by keeping a prompt short - only to spend the next ten minutes answering follow-up questions because the AI didn't have what it needed? Yeah, me too. I recently came across a really interesting post from the GitHub Copilot team about making ## AI-assisted customization may reshape how software is built DevFeed: [AI-assisted customization may reshape how software is built]() Original publisher: [Read original article]() Author: Mike Yerou Published: 2026-09-01T12:42:38Z Content type: opinion Language: en Sources: [Rock and Null]() Topics: [Software](), [Vibe coding](), [AI-assisted coding](), [Agent Harness](), [Cloudflare]() Tags: [agent-harness](), [ai](), [ai-coding-agents](), [cloudflare](), [software](), [software-engineering](), [thoughts](), [vibe-coding]() ### AI overview The article argues that AI coding agents and vibe coding are lowering the cost of custom software, but suggests the future may involve customizing capable software platforms rather than building and maintaining applications entirely from scratch. ### Source excerpt AI is changing the economics of custom software. The future may not be everyone building from scratch, but rather customizing a strong software base to fit their specific needs. ## Temporal Agent Harness: An early look at durable agent infrastructure DevFeed: [Temporal Agent Harness: An early look at durable agent infrastructure]() Original publisher: [Read original article]() Author: Cornelia Davis Published: 2026-08-20T00:00:00Z Content type: article Language: en Sources: [Temporal Blog]() Topics: [Agent Harness](), [AI Agent](), [reliability](), [interfaces]() Tags: [agent-harness](), [ai](), [ai-agents](), [announcements](), [execution](), [interfaces](), [production](), [tool]() ### AI overview Temporal presents an early look at Agent Harness, an infrastructure layer for production AI agents. It is designed to add durable execution, tool-call approvals, typed interfaces, enforceable policies, reliable execution, and a durable history of agent activity while allowing teams to keep using existing agent harnesses and tools. ### Source excerpt Temporal's new Agent Harness brings durable execution, tool-call approvals, and typed interfaces to production AI agents. An early look. ## Using the GitHub Copilot SDK for Java DevFeed: [Using the GitHub Copilot SDK for Java]() Original publisher: [Read original article]() Author: Edward Burns Published: 2026-08-10T19:30:00Z Content type: tutorial Language: en Sources: [GitHub]() Topics: [GitHub Copilot](), [Java](), [SDK](), [Jakarta EE](), [Agent Harness](), [Artificial Intelligence](), [Command-line interface](), [Spring AI]() Tags: [agent](), [agent-harness](), [ai](), [ai-ml](), [cli](), [client-library](), [engineering](), [framework](), [github-copilot](), [github-copilot-sdk](), [jakarta-ee](), [java](), [sdk]() ### AI overview This tutorial introduces the GitHub Copilot SDK for Java, a framework-agnostic client library for creating Copilot agent sessions, registering tools, sending prompts, and receiving structured responses. It demonstrates the SDK in a Jakarta EE 11 real-estate lead-management application and describes prerequisites including JDK, Maven, a GitHub Copilot subscription, and the Copilot CLI. ### Source excerpt Enterprise Java developers have a new superpower--drive GitHub Copilot from idiomatic Java code with annotations, virtual threads, and more. The post Using the GitHub Copilot SDK for Java appeared first on The GitHub Blog. ## What 21 five-minute demos looked like at WorkOS Demo Night DevFeed: [What 21 five-minute demos looked like at WorkOS Demo Night]() Original publisher: [Read original article]() Author: WorkOS Published: 2026-08-07T00:00:00Z Content type: article Language: en Sources: [WorkOS Blog]() Topics: [Demo](), [Agent Harness](), [Artificial Intelligence]() Tags: [agent](), [agent-harness](), [demo](), [mac](), [mac-os](), [recap]() ### AI overview A recap of WorkOS Demo Night in San Francisco, featuring 21 live five-minute demos. The article highlights Carbon Code, an agent harness running on emulated Mac OS 9 that edited its own source and added an effort selector during the event. ### Source excerpt A recap of the August 5 Demo Night at our San Francisco office: 21 live five-minute demos, no slides, no pitches, and an agent harness running on Mac OS 9. ## Build your own coding agent DevFeed: [Build your own coding agent]() Original publisher: [Read original article]() Author: Owain Lewis Published: 2026-08-05T20:06:19Z Content type: tutorial Language: en Sources: [The AI Engineer]() Topics: [Agent Harness](), [coding](), [AI-assisted coding](), [AI Development](), [Code](), [Go Language](), [Shell](), [Language models](), [Open Source](), [codex](), [Claude Code]() Tags: [agent](), [agent-harness](), [agents](), [ai](), [ai-development](), [claude](), [claude-code](), [code](), [codex](), [coding](), [commands](), [development](), [go](), [open-source]() ### AI overview This tutorial explains the basic architecture of coding agents and implements a minimal Go-based agent called Micro Neo. The agent connects to an existing AI model and supports reading files, editing files, and running shell commands. It also describes the harness loop that sends tasks and tool results between the user, model, and local machine. ### Source excerpt Why I built my own coding agent and lessons learned ## Data Engineering Weekly #281 DevFeed: [Data Engineering Weekly #281]() Original publisher: [Read original article]() Author: Ananth Packkildurai Published: 2026-08-03T12:34:40Z Content type: article Language: en Sources: [Data Engineering Weekly]() Topics: [data-engineering](), [AI, ML & Data Engineering](), [Loop Engineering](), [Artificial Intelligence](), [Large Language Model](), [post-training]() Tags: [agent-harness](), [ai](), [data-engineering](), [feature-engineering](), [genai](), [netflix](), [newsletter](), [post-training](), [weekly]() ### AI overview Data Engineering Weekly #281 covers building data platforms, emerging approaches to AI workflow architecture, data modernization, Netflix's GenRec recommendation system, AI infrastructure modernization, and evaluation practices for generative AI at scale. ### Source excerpt The Weekly Data Engineering Newsletter ## Testing an Agent Harness Without Ever Calling the Model DevFeed: [Testing an Agent Harness Without Ever Calling the Model]() Original publisher: [Read original article]() Author: Philipp Lehmann (philipp.lehmann@gruppe.ai) Published: 2026-07-31T07:00:00Z Content type: tutorial Language: en Sources: [Philipp Lehmann]() Topics: [Testing](), [Agent Harness](), [Pytest](), [Unit testing](), [coding]() Tags: [agent](), [agent-harness](), [agents](), [api](), [network](), [permission](), [pytest](), [python](), [recording](), [testing](), [unit-testing]() ### AI overview This tutorial explains how to unit test a coding agent's tool-call permission layer without calling a model. It recommends separating the decision function from model responses, using recorded real-request fixtures, and keeping the policy module independent of any LLM SDK. ### Source excerpt Unit testing a coding agent's tool-call permission layer with pytest and recorded fixtures: allow, deny or ask decisions with no model or API key. ## LoopSmith: Closed-Loop AI Engineering for Self-Correcting Pipelines on Antigravity 2.0 DevFeed: [LoopSmith: Closed-Loop AI Engineering for Self-Correcting Pipelines on Antigravity 2.0]() Original publisher: [Read original article]() Author: Esther Irawati Setiawan Published: 2026-07-29T09:40:30Z Content type: tutorial Language: en Sources: [Google Developer Experts - Medium]() Topics: [AI Engineering](), [google-antigravity](), [Agent Harness](), [SDKs](), [Command-line interface](), [Code]() Tags: [agent-harness](), [agentic-ai](), [ai](), [ai-engineering](), [antigravity](), [cli](), [google-antigravity](), [pipelines](), [sdk](), [verification]() ### AI overview A guide to using Antigravity 2.0 to build closed-loop AI engineering workflows. It presents a state-machine pattern in which an agent writes, runs, and fixes code against a defined objective until the output is verified, while noting that the SDK is pre-v1.0 and its documented interfaces may change. ### Source excerpt LoopSmith: Closed-Loop AI Engineering -- Autonomous /goal Execution for Self-Correcting Pipelines on Antigravity 2.0Stop prompting the agent turn by turn. Hand it an objective, a bar to clear, and let the state machine write, run, and fix its own code until the output is verified.This guide targets Antigravity 2.0 -- the four-surface release (desktop app, agy CLI, google-antigravity SDK, and enterprise cloud) that shares one agent harness. The SDK is pre-v1.0; symbol names and CLI flags below reflect the documented API as of mid-2026. Treat the patterns as stable and re-check exact signatures against the current docs before you ship.Table of contents The problem with the on-demand agent The architectural shift: closed-loop engineering as a state machine Step 1 -- Initialize the project and the state machine (agy) Step 2 -- Define the objective and trigger /goal execution mode (SDK) Step 3 -- Implement the self-correcting loop Step 4 -- Verification and state finalization Conclusion 1. The problem with the on-demand agent Most "AI engineering" today is still conversational. You prompt; the model answers. You notice the answer is wrong; you prompt again. You paste a traceback; it apologizes and tries once more. The intelligence is real -- but you are the control loop. You are the thing that runs the code, reads the error, decides whether the output is good enough, and feeds the next instruction back in. Take the human out of that seat and the whole system stops. That's fine for a chat window. It falls apart the moment you want an agent to produce a deliverable -- a cleaned dataset, a reconciled financial report, a migration that actually compiles. Real analytical work is iterative and self-referential: you write a script, it crashes on a currency string, you fix the parse, it runs but the totals don't reconcile, you fix the aggregation, and only then is the output trustworthy. Every one of those arrows is a decision. An on-demand agent makes you supply all of them. There are ## The Bare-Bones Coding Agent Loop DevFeed: [The Bare-Bones Coding Agent Loop]() Original publisher: [Read original article]() Author: Paul Iusztin Published: 2026-07-28T13:54:35Z Content type: tutorial Language: en Sources: [Decoding ML]() Topics: [Agent Harness](), [AI-assisted coding](), [Python](), [Shell](), [Text-based user interface]() Tags: [agent](), [agent-harness](), [coding](), [coding-agents](), [interfaces](), [python]() ### AI overview A lesson in an open-source course that explains how to build a bare-bones coding agent loop in Python. It covers decomposing goals into plans, executing them with tools, building a terminal user interface, and avoiding message-handling problems in the loop. ### Source excerpt One agent loop, 9 tools, and a terminal you can steer. ## OpenAI agent harness breached Hugging Face during a cybersecurity model evaluation DevFeed: [OpenAI agent harness breached Hugging Face during a cybersecurity model evaluation]() Original publisher: [Read original article]() Author: Simon Willison Published: 2026-07-22T23:51:33Z Content type: opinion Language: en Sources: [Simon Willison]() Topics: [Agent Harness](), [OpenAI](), [Exploit](), [LLM evaluation / benchmarking](), [Vulnerabilities](), [Cybersecurity](), [Large Language Model](), [GitHub](), [Linux Kernel](), [V8]() Tags: [accidental-cyberattacks](), [accidental-cyberattacks-15](), [agent-harness](), [ai](), [ai-2-235](), [ai-agents](), [ai-security-research](), [ai-security-research-42](), [anthropic](), [anthropic-336](), [attacks](), [benchmark](), [exploit](), [generative-ai](), [generative-ai-1-981](), [github](), [hugging-face](), [hugging-face-27](), [kernel](), [llm](), [llms](), [llms-1-947](), [openai](), [openai-463](), [openai-hugging-face-incident](), [openai-hugging-face-incident-9](), [paper-review](), [paper-review-19](), [research](), [sandboxing](), [sandboxing-55](), [security](), [security-634](), [vulnerabilities]() ### AI overview Simon Willison reviews a security incident in which an OpenAI agent harness, used during evaluation of an unreleased model with guardrails disabled, breached Hugging Face systems. The article also examines ExploitGym, a benchmark for testing whether LLM-powered agents can turn real-world vulnerability reports into concrete exploits. ### Source excerpt This story is wild. The short version: OpenAI were running a cybersecurity test against an unreleased model, with the model's guardrail features turned off. Rather than solve the test, the model broke its way out of OpenAI's sandbox, then found exploits to break in to Hugging Face, all so it could cheat on the test by stealing the answers. Along the way it helped make the strongest case yet for how the imbalance of model availability is hurting our ability to secure our software. Here's what happened We currently have three documents to help us understand what happened here. ExploitGym: Can AI Agents Turn Security Vulnerabilities into Real Attacks? is a paper published on 11th May 2026 describing ExploitGym, a new eval suite for LLM-powered agent systems. Security incident disclosure -- July 2026 by Hugging Face on 16th July 2026 describes how they detected an attack from an "agentic security-research harness - used LLM still not known" that breached some of their systems. OpenAI and Hugging Face partner to address security incident during model evaluation from OpenAI on 21st July 2026 confesses that it was their agent harness that did this, and that they're working with Hugging Face to clean up the mess. Update 5th August 2026: Hugging Face published a great deal more information about the attack on July 27th. ExploitGym I hadn't seen the ExploitGym paper before and it's a really interesting one. Authors from UC Berkeley, the Max Planck Institute, UC Santa Barbara, and Arizona State designed a new benchmark for evaluating models on their ability to turn a reported vulnerability into a concrete exploit. OpenAI, Anthropic, and Google provided feedback and helped run the benchmark against their models. The benchmark "comprises 898 instances derived from real-world vulnerabilities that affected popular software projects" - including the Linux kernel and V8 JavaScript engine. The ExploitGym benchmark is available on GitHub. Here's the paragraph that best represents their ## Building a Coding Agent From Scratch DevFeed: [Building a Coding Agent From Scratch]() Original publisher: [Read original article]() Author: Paul Iusztin Published: 2026-07-22T11:04:24Z Content type: tutorial Language: en Sources: [Decoding ML]() Topics: [Agent Harness](), [AI-assisted coding](), [coding](), [LangChain](), [Claude Code](), [codex]() Tags: [agent-harness](), [ai](), [ai-agents](), [ai-evals](), [benchmarks](), [claude-code](), [codex](), [coding-agents](), [langchain]() ### AI overview This tutorial explains how to build a coding-agent harness from scratch in Python. It covers the agent loop, shell execution, context engineering, subagents, remote parallel agents, and evaluation workflows, using the project Decode as the practical example. ### Source excerpt Designing the harness around the model, from the agent loop to a remote swarm. ## How Rapidflare built a million+ document ingestion pipeline for Agents on Temporal DevFeed: [How Rapidflare built a million+ document ingestion pipeline for Agents on Temporal]() Original publisher: [Read original article]() Author: Vasanth Asokan Published: 2026-07-08T00:00:00Z Content type: article Language: en Sources: [Temporal Blog]() Topics: [Agent Harness](), [Documentation](), [reliability](), [data](), [Graphs]() Tags: [agent-harness](), [agents](), [community](), [knowledge-graph](), [pipeline](), [pipelines](), [reliability](), [scale](), [technical]() ### AI overview Rapidflare explains how it uses Temporal to build durable, observable document-ingestion pipelines for technical sales agents processing knowledge bases at million-document scale. The article describes the need to ingest and pre-structure extensive technical literature into proprietary knowledge-graph formats, while preserving completeness and reliability. ### Source excerpt Rapidflare explains how it uses Temporal to run durable, observable document ingestion pipelines for technical sales Agents at million-document scale. ## Harness Bench: как оценить агентский harness и выбрать связку с моделью DevFeed: [Harness Bench: как оценить агентский harness и выбрать связку с моделью]() Original publisher: [Read original article]() Author: andrivasg (red\_mad\_robot) Published: 2026-06-30T11:30:03Z Content type: article Language: ru Sources: [Redmadrobot EN](), [Redmadrobot RU]() Topics: [Artificial Intelligence]() Tags: [agent-harness](), [ai](), [ai-e2239b5ae8fa](), [ai-evaluation](), [benchmark](), [harness](), [llm](), [mcp]() ### AI overview The article introduces Harness Bench, an open framework for evaluating model-and-harness combinations on real tasks under consistent conditions. It explains why an agent harness affects an AI system's practical capabilities, discusses problems in open-source harnesses that disrupt automated evaluation, and describes how existing benchmarks can be used for reproducible testing. ### Source excerpt Привет! Я Андрей Иванов, NLP-исследователь в R&D-лаборатории red_mad_robot. Когда мы собираем AI-агента, первым делом выбираем модель под задачу. Но в реальном приложении она не работает в одиночку, ей нужен агентский harness -- программная обвязка. Поэтому выбирать приходится не просто модель, а связку "модель + harness". Чтобы делать этот выбор осознанно, мы создали Harness Bench -- открытый фреймворк, который тестирует связки на реальных задачах в одинаковых условиях. В статье расскажу, как он устроен, разберу баги опенсорсных обвязок, которые ломают автоматический прогон, а потом покажу на цифрах, как смена harness влияет на способности одной и той же модели. Читать далее ## Loop Engineering: Designing Evaluated AI-Agent Workflows DevFeed: [Loop Engineering: Designing Evaluated AI-Agent Workflows]() Original publisher: [Read original article]() Author: Maxi Contieri Published: 2026-06-26T20:20:37Z Content type: tutorial Language: en Sources: [Maximiliano Contieri - Software Design]() Topics: [Loop Engineering](), [AI Agent](), [Agent Harness](), [Process]() Tags: [agent-harness](), [ai](), [build](), [engineering](), [evaluation](), [workflow]() ### AI overview This tutorial explains Loop Engineering as a workflow in which a system prompts an AI agent, rather than a person prompting it manually. It identifies four required parts--goal, context, evaluation, and agent--and emphasizes that evaluation provides exit criteria while a harness supplies operational boundaries. It also advocates starting with a narrowly scoped, testable specification and iterating from a failing evaluation. ### Source excerpt Everyone is talking about Loop Engineering. Apparently, you don't need to program anymore. TL;DR: Loop Engineering is the hottest AI workflow pattern of 2026. But it hides a dirty secret. The Tweet ## Using local Gemma and Qwen models to triage OpenClaw issues and pull requests DevFeed: [Using local Gemma and Qwen models to triage OpenClaw issues and pull requests]() Original publisher: [Read original article]() Author: Onur Solmaz; ben burtenshaw; shaun smith Published: 2026-06-22T00:00:00Z Content type: article Language: en Sources: [Hugging Face - Blog]() Topics: [OpenClaw](), [Agent Harness](), [Open Source](), [Artificial Intelligence](), [gemma](), [qwen]() Tags: [agent](), [agent-harness](), [artificial-intelligence](), [free](), [gemma](), [guide](), [open-source](), [open-source-collab](), [openclaw](), [pull-requests](), [qwen](), [real-time]() ### AI overview The article describes using local Gemma and Qwen models in an agent harness to classify and triage issues and pull requests in the OpenClaw repository. It presents local execution as a way to support near-real-time notifications without relying on a paid hosted-model quota, using structured outputs and a finite label set. ### Source excerpt We're on a journey to advance and democratize artificial intelligence through open source and open science. ## OpenCode power user tips DevFeed: [OpenCode power user tips]() Original publisher: [Read original article]() Author: Kaushik Gopal Published: 2026-06-03T16:00:00Z Content type: tutorial Language: en Sources: [Kaushik Gopal's Site]() Topics: [Agent Harness](), [Tooling](), [Open Source](), [Code]() Tags: [agent-harness](), [claude-code](), [codex](), [commands](), [fork](), [how-to](), [model](), [open-source](), [opencode](), [switching](), [tips]() ### AI overview A practical guide to advanced OpenCode features, including the leader key, session management, session forking, conversation rewind, and model switching. ### Source excerpt In this post, I'd like to talk about some power user tips for OpenCode - an open source, model agnostic harness that more people should be using. Hopefully some of the advanced use cases convince you to give OpenCode (and OpenChamber) a shot. intermediate to advanced tips only I am specifically choosing to talk about some advanced tips in this post. If you've never used an agent harness or are looking to learn how to use OpenCode, this post can be useful but reader beware. Unleash the leader key ## While ⌃p (Ctrl + P) will list out all the possible commands (and is helpful), OpenCode has the concept of a "leader" key (which defaults to ⌃x). The leader key allows you to execute targeted useful commands more quickly and there's a slew of useful ones pre-defined1. ctrl-p shows all commands. notice leader key bound to some Manage multiple sessions - leader l ## People reach for whole terminals and extra tooling to juggle between agent sessions. I too had an overly customized tmux setup that looked like this: Before: agent session listing via tmux OpenCode simplifies this. Just hit leader + l and you view current sessions and can instantly switch to that session by just selecting it from the list. After: leader-l allows quick session switching The ability to quickly rename a session from this view is a godsend for me and what lets me be organized. session directory filtering you can pass a --dir . flag to opencode when launching it, which filters the session list to just this workspace/directory by default. You can alternatively not pass that flag, and the session list will show all sessions. Fork (or Branch) sessions /fork ## Forking takes the session you're in and spawns a new one. You branch off into a separate conversation while the main agent keeps grinding on whatever you left it doing. I love this feature and even cobbled my own version with tmux long before most harnesses shipped it. Claude Code, Codex and other harnesses have caught up and support this feature. ## From prompt to production: Build full-stack apps faster with Google AI Studio and Firebase DevFeed: [From prompt to production: Build full-stack apps faster with Google AI Studio and Firebase]() Original publisher: [Read original article]() Author: Kara Yu; Sam Phillips Published: 2026-03-19T00:00:00Z Content type: release Language: en Sources: [Firebase Blog]() Topics: [Firebase](), [Google AI](), [Firestore](), [Google](), [Authentication](), [backends](), [Agent Harness]() Tags: [agent](), [agent-harness](), [ai](), [ai-studio](), [antigravity](), [authentication](), [backends](), [build](), [firebase](), [firebase-studio](), [firestore](), [google](), [security]() ### AI overview Firebase announces an integration with Google AI Studio that can provision Firebase projects, Firestore, Authentication, sign-in pages, application code, and draft Security Rules with user approval. The announcement also describes a new Google AI Studio coding agent for multi-step code edits and connections to services such as payment processors and Google Maps. ### Source excerpt Today, we're announcing that Firebase is now integrated with Google AI Studio, accelerating your path from prompt to production so you can turn your vibe-coded ideas into fully functional apps with robust backends. Read on to learn how this works and what this means for Firebase Studio. [Next page]()