<?xml version="1.0" encoding="utf-8"?>
<feed xmlns="http://www.w3.org/2005/Atom">
  <id>https://newruntime.com/feeds/coding-agents.xml</id>
  <title>New Runtime: Coding agents</title>
  <subtitle>A living map of coding agents as they move from autocomplete into supervised software work, review loops, environments, and operating cost. This bounded feed exists because the topic is a featured, evidence-dense New Runtime hub.</subtitle>
  <updated>2026-11-18T00:00:00.000Z</updated>
  <link rel="self" type="application/atom+xml" href="https://newruntime.com/feeds/coding-agents.xml"/>
  <link rel="hub" href="https://pubsubhubbub.appspot.com/"/>
  <link rel="alternate" type="text/html" href="https://newruntime.com/topics/coding-agents/"/>
  <author><name>New Runtime</name><uri>https://newruntime.com/</uri></author>
  <rights>New Runtime tracks the systems, signals, and shifts reshaping software as agents become first-class participants in work, products, and organizations.</rights>
  <entry>
    <id>https://newruntime.com/posts/designing-the-langfuse-ai-gateway-e28995/</id>
    <title>Designing the Langfuse AI Gateway</title>
    <link rel="alternate" href="https://newruntime.com/posts/designing-the-langfuse-ai-gateway-e28995/"/>
    <published>2026-11-18T00:00:00.000Z</published>
    <updated>2026-11-18T00:00:00.000Z</updated>
    <summary>Langfuse explains how it designed an AI Gateway for coding-agent traffic, choosing Rust and Tokio to handle large requests, long-lived streams, and bursty CPU work.</summary>
    <category term="ai-infrastructure"/>
    <category term="coding-agents"/>
    <category term="inference"/>
    <link rel="related" href="https://langfuse.com/blog/2026-11-11-building-the-langfuse-ai-gateway"/>
  </entry>
  <entry>
    <id>https://newruntime.com/posts/factory-expands-collaboration-with-microsoft-to-bring-autonomous-6537fe/</id>
    <title>Factory Expands Collaboration with Microsoft to Bring Autonomous Software Engineering to Enterprises</title>
    <link rel="alternate" href="https://newruntime.com/posts/factory-expands-collaboration-with-microsoft-to-bring-autonomous-6537fe/"/>
    <published>2026-11-18T00:00:00.000Z</published>
    <updated>2026-11-18T00:00:00.000Z</updated>
    <summary>Factory says enterprises can run its Droids inside their Azure environments, connecting agent-driven engineering work to existing identity, network, source-control and approval systems.</summary>
    <category term="coding-agents"/>
    <category term="enterprise-agents"/>
    <category term="enterprise-ai"/>
    <link rel="related" href="https://factory.com/news/microsoft-collaboration"/>
  </entry>
  <entry>
    <id>https://newruntime.com/posts/coding-agent-meet-ngrok-the-ngrok-mcp-server-7f32e5/</id>
    <title>Coding agent, meet ngrok: the ngrok MCP server is live</title>
    <link rel="alternate" href="https://newruntime.com/posts/coding-agent-meet-ngrok-the-ngrok-mcp-server-7f32e5/"/>
    <published>2026-11-17T00:00:00.000Z</published>
    <updated>2026-11-17T00:00:00.000Z</updated>
    <summary>ngrok launched a generally available MCP server that lets coding agents inspect account resources and traffic, consult current documentation, and manage cloud endpoint Traffic Policy.</summary>
    <category term="coding-agents"/>
    <category term="mcp"/>
    <link rel="related" href="https://ngrok.com/blog/mcp"/>
  </entry>
  <entry>
    <id>https://newruntime.com/posts/autoswarm-from-hku-orchestrates-multiple-commercial-coding-agents-9b9f00/</id>
    <title>AutoSwarm from HKU orchestrates multiple commercial coding agents as one coordinated team.</title>
    <link rel="alternate" href="https://newruntime.com/posts/autoswarm-from-hku-orchestrates-multiple-commercial-coding-agents-9b9f00/"/>
    <published>2026-11-16T00:00:00.000Z</published>
    <updated>2026-11-16T00:00:00.000Z</updated>
    <summary>HKUDS’s AutoSwarm is an MIT-licensed tool for coordinating existing AI agents through a shared workflow, including coding agents such as Codex, Claude Code, OpenCode, and Cursor.</summary>
    <category term="claude"/>
    <category term="claude-code"/>
    <category term="coding-agents"/>
    <link rel="related" href="https://github.com/HKUDS/AutoSwarm"/>
  </entry>
  <entry>
    <id>https://newruntime.com/posts/browseros-open-sources-neo-a-browser-for-coding-71629c/</id>
    <title>BrowserOS open-sources &apos;neo&apos;, a browser for coding agents with per-agent tabs and session replays</title>
    <link rel="alternate" href="https://newruntime.com/posts/browseros-open-sources-neo-a-browser-for-coding-71629c/"/>
    <published>2026-11-16T00:00:00.000Z</published>
    <updated>2026-11-16T00:00:00.000Z</updated>
    <summary>BrowserOS has open-sourced neo, a browser environment intended for coding agents.</summary>
    <category term="claude"/>
    <category term="claude-code"/>
    <category term="coding-agents"/>
    <link rel="related" href="https://github.com/browseros-ai/BrowserOS"/>
  </entry>
  <entry>
    <id>https://newruntime.com/posts/how-we-make-together-ai-easy-for-coding-181c45/</id>
    <title>How we make Together AI easy for coding agents to use</title>
    <link rel="alternate" href="https://newruntime.com/posts/how-we-make-together-ai-easy-for-coding-181c45/"/>
    <published>2026-11-16T00:00:00.000Z</published>
    <updated>2026-11-16T00:00:00.000Z</updated>
    <summary>Together AI says small changes to its documentation, agent skills, and API error handling raised coding-agent task completion from 75% to 97% in its evaluations.</summary>
    <category term="coding-agents"/>
    <category term="skills"/>
    <link rel="related" href="https://www.together.ai/blog/how-we-do-agent-evals-at-together-ai"/>
  </entry>
  <entry>
    <id>https://newruntime.com/posts/clamdown-keep-coding-agents-working-on-a-macbook-337b4b/</id>
    <title>ClamDown: Keep coding agents working on a MacBook with the lid closed</title>
    <link rel="alternate" href="https://newruntime.com/posts/clamdown-keep-coding-agents-working-on-a-macbook-337b4b/"/>
    <published>2026-11-15T00:00:00.000Z</published>
    <updated>2026-11-15T00:00:00.000Z</updated>
    <summary>ClamDown is a free, open-source macOS menu bar app that keeps a MacBook awake with its lid closed while managing fan speed, screen locking, battery and heat cut-offs, and recovery if the app stops responding.</summary>
    <category term="claude"/>
    <category term="claude-code"/>
    <category term="coding-agents"/>
    <link rel="related" href="https://github.com/delef/ClamDown"/>
  </entry>
  <entry>
    <id>https://newruntime.com/posts/artificial-analysis-publishes-coding-agent-index-v2-1-983801/</id>
    <title>Artificial Analysis publishes Coding Agent Index v2.1 with FrontierCode-AA 2.0, Terminal-Bench 4.1 and SWE-Atlas-QnA-AA</title>
    <link rel="alternate" href="https://newruntime.com/posts/artificial-analysis-publishes-coding-agent-index-v2-1-983801/"/>
    <published>2026-11-14T00:00:00.000Z</published>
    <updated>2026-11-14T00:00:00.000Z</updated>
    <summary>Artificial Analysis has updated its Coding Agent Index to v2.1, combining three benchmarks for software engineering, terminal use and technical Q&amp;A.</summary>
    <category term="agents"/>
    <category term="coding-agents"/>
    <category term="evals"/>
    <link rel="related" href="https://artificialanalysis.ai/agents/coding"/>
  </entry>
  <entry>
    <id>https://newruntime.com/posts/caveman-is-an-open-source-cli-that-changes-23b835/</id>
    <title>Caveman is an open-source CLI that changes how coding agents handle responses, file output</title>
    <link rel="alternate" href="https://newruntime.com/posts/caveman-is-an-open-source-cli-that-changes-23b835/"/>
    <published>2026-11-14T00:00:00.000Z</published>
    <updated>2026-11-14T00:00:00.000Z</updated>
    <summary>Caveman is an open-source CLI that changes how coding agents handle responses, file output, and repeated scripts, with the stated aim of reducing token use and cost.</summary>
    <category term="claude"/>
    <category term="claude-code"/>
    <category term="coding-agents"/>
    <link rel="related" href="https://github.com/JuliusBrussee/caveman"/>
  </entry>
  <entry>
    <id>https://newruntime.com/posts/gumloop-says-its-internal-software-factory-merged-571-68e712/</id>
    <title>Gumloop says its internal Software Factory merged 571 agent-written pull requests in six weeks</title>
    <link rel="alternate" href="https://newruntime.com/posts/gumloop-says-its-internal-software-factory-merged-571-68e712/"/>
    <published>2026-11-13T00:00:00.000Z</published>
    <updated>2026-11-13T00:00:00.000Z</updated>
    <summary>Gumloop says its internal Software Factory merged 571 agent-written pull requests in six weeks, while engineers retained review and merge authority.</summary>
    <category term="automation"/>
    <category term="code-review"/>
    <category term="coding-agents"/>
    <link rel="related" href="https://www.gumloop.com/engineering/software-factory"/>
  </entry>
  <entry>
    <id>https://newruntime.com/posts/how-to-train-a-transformer-as-a-software-32d1f6/</id>
    <title>How to train a Transformer (as a software engineer with no ML experience)</title>
    <link rel="alternate" href="https://newruntime.com/posts/how-to-train-a-transformer-as-a-software-32d1f6/"/>
    <published>2026-11-13T00:00:00.000Z</published>
    <updated>2026-11-13T00:00:00.000Z</updated>
    <summary>A software engineer with no prior neural-network training experience built a transformer to judge positions in a CodinGame bot, using coding agents to write the model and training code.</summary>
    <category term="claude"/>
    <category term="coding-agents"/>
    <link rel="related" href="https://aiengg.dev/blog/how-to-train-a-transformer"/>
  </entry>
  <entry>
    <id>https://newruntime.com/posts/recursive-model-card-reports-flint-s-89-7-cdd01f/</id>
    <title>Recursive model card reports Flint’s 89.7% SWE-bench Pro v2 score</title>
    <link rel="alternate" href="https://newruntime.com/posts/recursive-model-card-reports-flint-s-89-7-cdd01f/"/>
    <published>2026-11-13T00:00:00.000Z</published>
    <updated>2026-11-13T00:00:00.000Z</updated>
    <summary>Recursive’s Flint model card reports 89.7% on SWE-bench Pro v2, alongside scores on SWE-bench Verified and Terminal-Bench 2.1.</summary>
    <category term="coding-agents"/>
    <category term="evals"/>
    <category term="models"/>
    <link rel="related" href="https://huggingface.co/RecursiveSI/flint/blob/main/README.md"/>
  </entry>
  <entry>
    <id>https://newruntime.com/posts/ace-github-next-s-open-source-tool-lets-3659bb/</id>
    <title>Ace: GitHub Next&apos;s open-source tool lets teams and agents collaborate on the same Mac.</title>
    <link rel="alternate" href="https://newruntime.com/posts/ace-github-next-s-open-source-tool-lets-3659bb/"/>
    <published>2026-11-12T00:00:00.000Z</published>
    <updated>2026-11-12T00:00:00.000Z</updated>
    <summary>GitHub Next has released Ace, an experimental open-source tool for teams to collaborate with coding agents running on their own machines.</summary>
    <category term="anthropic"/>
    <category term="coding-agents"/>
    <category term="github"/>
    <link rel="related" href="https://ace.githubnext.com/"/>
  </entry>
  <entry>
    <id>https://newruntime.com/posts/gemini-4-hydrogen-now-available-on-ai-gateway-7b3652/</id>
    <title>Gemini 4 Hydrogen now available on AI Gateway</title>
    <link rel="alternate" href="https://newruntime.com/posts/gemini-4-hydrogen-now-available-on-ai-gateway-7b3652/"/>
    <published>2026-11-12T00:00:00.000Z</published>
    <updated>2026-11-12T00:00:00.000Z</updated>
    <summary>Vercel says Gemini 4 Hydrogen is now available through AI Gateway, where developers can call it as google/gemini-4-hydrogen and configure it in several coding agents.</summary>
    <category term="claude"/>
    <category term="coding-agents"/>
    <category term="gemini"/>
    <link rel="related" href="https://vercel.com/changelog/gemini-4-hydrogen-now-available-on-ai-gateway"/>
  </entry>
  <entry>
    <id>https://newruntime.com/posts/microsoft-publishes-two-studies-on-post-training-supply-edd99e/</id>
    <title>Microsoft publishes two studies on post-training supply chain backdoors in coding agents</title>
    <link rel="alternate" href="https://newruntime.com/posts/microsoft-publishes-two-studies-on-post-training-supply-edd99e/"/>
    <published>2026-11-12T00:00:00.000Z</published>
    <updated>2026-11-12T00:00:00.000Z</updated>
    <summary>Microsoft Security Research and the Microsoft AI Red Team describe two ways post-training artifacts can carry dormant backdoors into coding agents: poisoned fine-tuning data and attacker-trained model checkpoints.</summary>
    <category term="claude"/>
    <category term="coding-agents"/>
    <link rel="related" href="https://www.microsoft.com/en-us/security/blog/2026/11/12/trained-to-betray-how-post-training-supply-chain-attacks-hijack-coding-agents"/>
  </entry>
  <entry>
    <id>https://newruntime.com/posts/chrome-releases-modern-web-guidance-1-0-with-977b2e/</id>
    <title>Chrome releases Modern Web Guidance 1.0 with Firefox and Edge joining the project</title>
    <link rel="alternate" href="https://newruntime.com/posts/chrome-releases-modern-web-guidance-1-0-with-977b2e/"/>
    <published>2026-11-11T00:00:00.000Z</published>
    <updated>2026-11-11T00:00:00.000Z</updated>
    <summary>Google says Modern Web Guidance 1.0 is now stable, with a workflow that scans existing web projects for selected modernization opportunities and guides coding agents through changes.</summary>
    <category term="coding-agents"/>
    <category term="google"/>
    <link rel="related" href="https://developer.chrome.com/blog/modern-web-guidance-v1"/>
  </entry>
  <entry>
    <id>https://newruntime.com/posts/google-chrome-s-modern-web-guidance-packages-expert-727856/</id>
    <title>Google Chrome’s Modern Web Guidance packages expert-curated web-platform recommendations as skills that</title>
    <link rel="alternate" href="https://newruntime.com/posts/google-chrome-s-modern-web-guidance-packages-expert-727856/"/>
    <published>2026-11-11T00:00:00.000Z</published>
    <updated>2026-11-11T00:00:00.000Z</updated>
    <summary>Google Chrome’s Modern Web Guidance packages expert-curated web-platform recommendations as skills that coding agents can retrieve through a CLI.</summary>
    <category term="coding-agents"/>
    <category term="google"/>
    <link rel="related" href="https://github.com/GoogleChrome/modern-web-guidance-src"/>
  </entry>
  <entry>
    <id>https://newruntime.com/posts/orchestrating-a-one-person-software-factory-run-many-5705c0/</id>
    <title>Orchestrating a One-Person Software Factory: run many coding agents without becoming their PM</title>
    <link rel="alternate" href="https://newruntime.com/posts/orchestrating-a-one-person-software-factory-run-many-5705c0/"/>
    <published>2026-11-11T00:00:00.000Z</published>
    <updated>2026-11-11T00:00:00.000Z</updated>
    <summary>Oskar Groth describes how he coordinates coding agents across a one-person software company: lane-level orchestrators delegate to agents that own discrete outcomes, while GitHub issues and pull requests provide the shared record and review gate.</summary>
    <category term="claude"/>
    <category term="coding-agents"/>
    <category term="github"/>
    <link rel="related" href="https://oskargroth.com/blog/orchestrating-agents"/>
  </entry>
  <entry>
    <id>https://newruntime.com/posts/berkeley-evolving-agent-teams-supervisor-restructured-coding-agents-4a7472/</id>
    <title>Berkeley &apos;Evolving Agent Teams&apos;: supervisor-restructured coding agents beat FlashInfer MLA decode kernel by 1.21x</title>
    <link rel="alternate" href="https://newruntime.com/posts/berkeley-evolving-agent-teams-supervisor-restructured-coding-agents-4a7472/"/>
    <published>2026-11-10T00:00:00.000Z</published>
    <updated>2026-11-10T00:00:00.000Z</updated>
    <summary>A Berkeley research team introduced Evolving Agent Teams, a framework that lets a supervisor restructure coding agents when they stall on GPU-kernel optimization.</summary>
    <category term="coding-agents"/>
    <link rel="related" href="https://arxiv.org/abs/2610.36848"/>
  </entry>
  <entry>
    <id>https://newruntime.com/posts/how-we-rebuilt-the-shopify-app-in-12-c85550/</id>
    <title>How we rebuilt the Shopify app in 12 weeks</title>
    <link rel="alternate" href="https://newruntime.com/posts/how-we-rebuilt-the-shopify-app-in-12-c85550/"/>
    <published>2026-11-10T00:00:00.000Z</published>
    <updated>2026-11-10T00:00:00.000Z</updated>
    <summary>Shopify says about 30 engineers rebuilt its mobile app from React Native to Swift and Kotlin in 12 weeks, with agents writing most of the code.</summary>
    <category term="agents"/>
    <category term="coding-agents"/>
    <link rel="related" href="https://shopify.engineering/shopify-app-rebuild"/>
  </entry>
  <entry>
    <id>https://newruntime.com/posts/particle-1-is-a-coding-model-trained-to-fe301e/</id>
    <title>Particle-1 is a coding model trained to finish tasks with fewer tokens</title>
    <link rel="alternate" href="https://newruntime.com/posts/particle-1-is-a-coding-model-trained-to-fe301e/"/>
    <published>2026-11-10T00:00:00.000Z</published>
    <updated>2026-11-10T00:00:00.000Z</updated>
    <summary>Particle AI’s Particle-1 is listed on OpenRouter as an agentic coding model post-trained on DeepSeek V4.1 Flash, with API access, listed token prices, and a one-million-token context window.</summary>
    <category term="coding-agents"/>
    <category term="models"/>
    <category term="pricing"/>
    <link rel="related" href="https://openrouter.ai/particle-ai/particle-1"/>
  </entry>
  <entry>
    <id>https://newruntime.com/signals/agents-now-build-3d-scenes-and-physical-designs-as-code-ce016f/</id>
    <title>Agents now build 3D scenes and physical designs as code</title>
    <link rel="alternate" href="https://newruntime.com/signals/agents-now-build-3d-scenes-and-physical-designs-as-code-ce016f/"/>
    <published>2026-11-09T00:00:00.000Z</published>
    <updated>2026-11-09T00:00:00.000Z</updated>
    <summary>Voyager-so&apos;s vork is a headless engine where coding agents write 3D scenes as code and render them on a GPU.</summary>
    <category term="coding-agents"/>
    <link rel="related" href="https://github.com/voyager-so/vork"/>
    <link rel="related" href="https://github.com/Sandflow-ai/sand3d"/>
    <link rel="related" href="https://standardphysics.com/archimedes"/>
    <link rel="related" href="https://standardphysics.com/reynolds"/>
  </entry>
  <entry>
    <id>https://newruntime.com/posts/inferencex-measures-muse-spark-open-on-gb300-nvl72-e05cc5/</id>
    <title>InferenceX measures Muse Spark Open on GB300 NVL72 at 43,643 tok/s per GPU and $0.015/M tokens</title>
    <link rel="alternate" href="https://newruntime.com/posts/inferencex-measures-muse-spark-open-on-gb300-nvl72-e05cc5/"/>
    <published>2026-11-09T00:00:00.000Z</published>
    <updated>2026-11-09T00:00:00.000Z</updated>
    <summary>SemiAnalysis InferenceX ranks GPUs serving Meta’s Muse Spark Open on an agentic coding workload, measuring throughput at the same per-user interaction target.</summary>
    <category term="coding-agents"/>
    <category term="compute"/>
    <category term="inference"/>
    <link rel="related" href="https://inferencex.semianalysis.com/rankings/fastest-gpu-for-muse-spark-open"/>
  </entry>
  <entry>
    <id>https://newruntime.com/posts/print3d-the-best-skill-for-3d-printing-071dd7/</id>
    <title>print3d: The best skill for 3D printing</title>
    <link rel="alternate" href="https://newruntime.com/posts/print3d-the-best-skill-for-3d-printing-071dd7/"/>
    <published>2026-11-09T00:00:00.000Z</published>
    <updated>2026-11-09T00:00:00.000Z</updated>
    <summary>The open-source print3d project connects coding agents to a workflow for turning a prompt, photo or shop listing into a checked, sliced 3D-print project.</summary>
    <category term="claude"/>
    <category term="claude-code"/>
    <category term="coding-agents"/>
    <link rel="related" href="https://github.com/mfranzon/print3d"/>
  </entry>
  <entry>
    <id>https://newruntime.com/posts/pylon-x-devin-from-bug-report-to-code-a5809f/</id>
    <title>Pylon x Devin: From bug report to code fix</title>
    <link rel="alternate" href="https://newruntime.com/posts/pylon-x-devin-from-bug-report-to-code-a5809f/"/>
    <published>2026-11-09T00:00:00.000Z</published>
    <updated>2026-11-09T00:00:00.000Z</updated>
    <summary>Pylon says its integration with Cognition’s Devin can turn a customer-reported bug into a coding session carrying the ticket context, then return session and pull-request updates to Pylon.</summary>
    <category term="agent-workflows"/>
    <category term="coding-agents"/>
    <link rel="related" href="https://www.usepylon.com/blog/pylon-x-devin"/>
  </entry>
  <entry>
    <id>https://newruntime.com/posts/vork-is-a-headless-rust-slang-3d-engine-12ecec/</id>
    <title>vork is a headless Rust/Slang 3D engine where coding agents write scenes as code and render on GPU.</title>
    <link rel="alternate" href="https://newruntime.com/posts/vork-is-a-headless-rust-slang-3d-engine-12ecec/"/>
    <published>2026-11-09T00:00:00.000Z</published>
    <updated>2026-11-09T00:00:00.000Z</updated>
    <summary>Voyager-so’s GitHub project describes vork as a headless 3D and VFX engine for agents, with scenes written as code and rendered on a GPU.</summary>
    <category term="coding-agents"/>
    <category term="github"/>
    <link rel="related" href="https://github.com/voyager-so/vork"/>
  </entry>
  <entry>
    <id>https://newruntime.com/posts/continuity-research-released-tess-4-5-27b-a-1c8d16/</id>
    <title>Continuity Research released Tess-4.5-27B, a 27B Apache-2.0 fine-tune of Qwen3.8-27B for agentic coding.</title>
    <link rel="alternate" href="https://newruntime.com/posts/continuity-research-released-tess-4-5-27b-a-1c8d16/"/>
    <published>2026-11-08T00:00:00.000Z</published>
    <updated>2026-11-08T00:00:00.000Z</updated>
    <summary>Continuity Research released Tess-4.5-27B, an open-weights, 27-billion-parameter fine-tune of Qwen3.8-27B aimed at engineering and agentic coding work.</summary>
    <category term="coding-agents"/>
    <category term="models"/>
    <category term="open-models"/>
    <link rel="related" href="https://huggingface.co/migtissera/Tess-4.5-27B"/>
  </entry>
  <entry>
    <id>https://newruntime.com/posts/edith-v1-0-0-open-source-native-macos-0268b3/</id>
    <title>Edith v1.0.0: open-source native macOS control center that coding agents can drive over MCP</title>
    <link rel="alternate" href="https://newruntime.com/posts/edith-v1-0-0-open-source-native-macos-0268b3/"/>
    <published>2026-11-08T00:00:00.000Z</published>
    <updated>2026-11-08T00:00:00.000Z</updated>
    <summary>Edith v1.0.0 is now the latest tagged release of the pulkitxm project.</summary>
    <category term="coding-agents"/>
    <category term="github"/>
    <category term="mcp"/>
    <link rel="related" href="https://github.com/pulkitxm/edith/releases/tag/v1.0.0"/>
  </entry>
  <entry>
    <id>https://newruntime.com/posts/openqodex-0-15-0-adds-java-support-to-dbf796/</id>
    <title>OpenQodex 0.15.0 adds Java support to its code graph</title>
    <link rel="alternate" href="https://newruntime.com/posts/openqodex-0-15-0-adds-java-support-to-dbf796/"/>
    <published>2026-11-08T00:00:00.000Z</published>
    <updated>2026-11-08T00:00:00.000Z</updated>
    <summary>OpenQodex 0.15.0 adds Java parsing to its code graph without requiring a build step.</summary>
    <category term="coding-agents"/>
    <category term="developer-tools"/>
    <link rel="related" href="https://github.com/openqodex/openqodex/releases/tag/openqodex@0.15.0"/>
  </entry>
  <entry>
    <id>https://newruntime.com/posts/openqodex-0-16-0-links-junit-tests-to-e2e2f5/</id>
    <title>OpenQodex 0.16.0 links JUnit tests to Java code and keeps possible issues out of verdicts</title>
    <link rel="alternate" href="https://newruntime.com/posts/openqodex-0-16-0-links-junit-tests-to-e2e2f5/"/>
    <published>2026-11-08T00:00:00.000Z</published>
    <updated>2026-11-08T00:00:00.000Z</updated>
    <summary>OpenQodex 0.16.0 adds links from JUnit 4 and JUnit 5 test methods to the Java code they call, alongside broader Java code-graph work.</summary>
    <category term="coding-agents"/>
    <category term="developer-tools"/>
    <link rel="related" href="https://github.com/openqodex/openqodex/releases"/>
  </entry>
  <entry>
    <id>https://newruntime.com/posts/testerarmy-is-now-e2e-announcing-our-8-million-37bb99/</id>
    <title>TesterArmy is now e2e. Announcing our $8 million seed round.</title>
    <link rel="alternate" href="https://newruntime.com/posts/testerarmy-is-now-e2e-announcing-our-8-million-37bb99/"/>
    <published>2026-11-08T00:00:00.000Z</published>
    <updated>2026-11-08T00:00:00.000Z</updated>
    <summary>TesterArmy has renamed itself e2e and announced an $8 million seed round to build what it calls a software verification layer for coding agents.</summary>
    <category term="coding-agents"/>
    <link rel="related" href="https://e2e.com/blog/seed"/>
  </entry>
  <entry>
    <id>https://newruntime.com/posts/the-missing-semester-of-your-cs-education-9c5745/</id>
    <title>The Missing Semester of Your CS Education</title>
    <link rel="alternate" href="https://newruntime.com/posts/the-missing-semester-of-your-cs-education-9c5745/"/>
    <published>2026-11-08T00:00:00.000Z</published>
    <updated>2026-11-08T00:00:00.000Z</updated>
    <summary>MIT’s 2026 edition of The Missing Semester puts agentic coding on the syllabus and says it has folded applicable AI tools and techniques into lectures across the course.</summary>
    <category term="coding-agents"/>
    <category term="developer-tools"/>
    <link rel="related" href="https://missing.csail.mit.edu/"/>
  </entry>
  <entry>
    <id>https://newruntime.com/posts/wakevane-open-source-event-driven-harness-for-long-c81475/</id>
    <title>WakeVane: open-source, event-driven harness for long-running coding agents, built on Databricks&apos; Omnigent (alpha)</title>
    <link rel="alternate" href="https://newruntime.com/posts/wakevane-open-source-event-driven-harness-for-long-c81475/"/>
    <published>2026-11-08T00:00:00.000Z</published>
    <updated>2026-11-08T00:00:00.000Z</updated>
    <summary>WakeVane is an Apache-2.0 open-source distribution of Databricks’ Omnigent, currently in alpha, that wakes coding agents on repository events and runs them in cloud sandboxes.</summary>
    <category term="claude"/>
    <category term="claude-code"/>
    <category term="coding-agents"/>
    <link rel="related" href="https://www.wakevane.com/"/>
  </entry>
  <entry>
    <id>https://newruntime.com/posts/accepted-but-wrong-typical-input-gates-release-41-75c83c/</id>
    <title>Accepted but Wrong: typical-input gates release 41% of wrong LLM API-migration repairs</title>
    <link rel="alternate" href="https://newruntime.com/posts/accepted-but-wrong-typical-input-gates-release-41-75c83c/"/>
    <published>2026-11-07T00:00:00.000Z</published>
    <updated>2026-11-07T00:00:00.000Z</updated>
    <summary>A preregistered study of 73 API and runtime migration tasks found that ordinary tests on typical inputs released 41% of incorrect LLM repairs.</summary>
    <category term="coding-agents"/>
    <category term="verification"/>
    <link rel="related" href="https://arxiv.org/abs/2611.08269"/>
  </entry>
  <entry>
    <id>https://newruntime.com/signals/agent-interfaces-are-becoming-the-way-tools-expose-capabilities-across-b9c619/</id>
    <title>Agent interfaces are becoming the way tools expose capabilities across products</title>
    <link rel="alternate" href="https://newruntime.com/signals/agent-interfaces-are-becoming-the-way-tools-expose-capabilities-across-b9c619/"/>
    <published>2026-11-07T00:00:00.000Z</published>
    <updated>2026-11-07T00:00:00.000Z</updated>
    <summary>Lineage uses MCP to give coding agents access to cross-repository code, agreed API contracts and change-request history, while Poki uses MCP to let them operate a physical iPhone over USB.</summary>
    <category term="agent-interfaces"/>
    <category term="coding-agents"/>
    <category term="mcp"/>
    <link rel="related" href="https://github.com/tanwarAalok/lineage"/>
    <link rel="related" href="https://github.com/JosssphZhou/poki"/>
  </entry>
  <entry>
    <id>https://newruntime.com/posts/google-says-stale-inline-bug-comments-can-divert-668466/</id>
    <title>Google says stale inline bug comments can divert coding agents into work nobody requested</title>
    <link rel="alternate" href="https://newruntime.com/posts/google-says-stale-inline-bug-comments-can-divert-668466/"/>
    <published>2026-11-07T00:00:00.000Z</published>
    <updated>2026-11-07T00:00:00.000Z</updated>
    <summary>Google says stale inline bug comments can divert coding agents into work nobody requested.</summary>
    <category term="coding-agents"/>
    <category term="google"/>
    <link rel="related" href="https://developers.googleblog.com/agent-distraction"/>
  </entry>
  <entry>
    <id>https://newruntime.com/posts/lineage-open-sources-an-mcp-server-that-gives-8d644f/</id>
    <title>Lineage open-sources an MCP server that gives coding agents cross-repo code and CR history</title>
    <link rel="alternate" href="https://newruntime.com/posts/lineage-open-sources-an-mcp-server-that-gives-8d644f/"/>
    <published>2026-11-07T00:00:00.000Z</published>
    <updated>2026-11-07T00:00:00.000Z</updated>
    <summary>Lineage is an open-source MCP server that lets coding agents search code across a team’s repositories and retrieve approved rules, API contracts and change-request history.</summary>
    <category term="coding-agents"/>
    <category term="github"/>
    <category term="mcp"/>
    <link rel="related" href="https://github.com/tanwarAalok/lineage"/>
  </entry>
  <entry>
    <id>https://newruntime.com/posts/openrouter-launches-auto-code-router-for-coding-agents-80b0b2/</id>
    <title>OpenRouter launches Auto Code router for coding agents</title>
    <link rel="alternate" href="https://newruntime.com/posts/openrouter-launches-auto-code-router-for-coding-agents-80b0b2/"/>
    <published>2026-11-07T00:00:00.000Z</published>
    <updated>2026-11-07T00:00:00.000Z</updated>
    <summary>OpenRouter has launched Auto Code, a model router designed for coding agents, alongside a rebuilt general-purpose Auto Router.</summary>
    <category term="claude"/>
    <category term="coding-agents"/>
    <link rel="related" href="https://openrouter.ai/blog/announcements/auto-and-auto-code-routers"/>
  </entry>
  <entry>
    <id>https://newruntime.com/signals/ordinary-text-in-an-agent-s-own-context-is-now-a-source-of-unauthorize-b39670/</id>
    <title>Ordinary text in an agent&apos;s own context is now a source of unauthorized actions</title>
    <link rel="alternate" href="https://newruntime.com/signals/ordinary-text-in-an-agent-s-own-context-is-now-a-source-of-unauthorize-b39670/"/>
    <published>2026-11-07T00:00:00.000Z</published>
    <updated>2026-11-07T00:00:00.000Z</updated>
    <summary>Google found that stale inline bug comments lead coding agents into work nobody asked for, and most of the resulting bug-related actions were unprompted.</summary>
    <category term="coding-agents"/>
    <category term="google"/>
    <link rel="related" href="https://developers.googleblog.com/agent-distraction"/>
    <link rel="related" href="https://arxiv.org/abs/2611.06545"/>
  </entry>
  <entry>
    <id>https://newruntime.com/posts/poki-open-sources-an-mcp-server-that-lets-a0f3aa/</id>
    <title>Poki open-sources an MCP server that lets coding agents drive a real iPhone over USB</title>
    <link rel="alternate" href="https://newruntime.com/posts/poki-open-sources-an-mcp-server-that-lets-a0f3aa/"/>
    <published>2026-11-07T00:00:00.000Z</published>
    <updated>2026-11-07T00:00:00.000Z</updated>
    <summary>Poki is an open-source MCP server and optional terminal agent for operating a real iPhone from a Mac, with no jailbreak or extra hardware.</summary>
    <category term="claude"/>
    <category term="coding-agents"/>
    <category term="mcp"/>
    <link rel="related" href="https://github.com/JosssphZhou/poki"/>
  </entry>
  <entry>
    <id>https://newruntime.com/posts/qoder-offers-free-daily-qoduck-stealth-model-for-cf50b0/</id>
    <title>Qoder offers free daily Qoduck stealth model for individual users</title>
    <link rel="alternate" href="https://newruntime.com/posts/qoder-offers-free-daily-qoduck-stealth-model-for-cf50b0/"/>
    <published>2026-11-07T00:00:00.000Z</published>
    <updated>2026-11-07T00:00:00.000Z</updated>
    <summary>Qoder says individual Qoder International users will receive a daily free allowance for Qoduck beginning November 7, 2026.</summary>
    <category term="coding-agents"/>
    <category term="models"/>
    <link rel="related" href="https://docs.qoder.com/events/qoduck-free"/>
  </entry>
  <entry>
    <id>https://newruntime.com/posts/step-zero-attack-paper-coding-agents-can-exhaust-a26476/</id>
    <title>Step-Zero Attack paper: coding agents can exhaust disk before model sees content</title>
    <link rel="alternate" href="https://newruntime.com/posts/step-zero-attack-paper-coding-agents-can-exhaust-a26476/"/>
    <published>2026-11-07T00:00:00.000Z</published>
    <updated>2026-11-07T00:00:00.000Z</updated>
    <summary>A paper on arXiv reports that coding agents can exceed disk limits while processing crafted archives before their models see any contents.</summary>
    <category term="coding-agents"/>
    <link rel="related" href="https://arxiv.org/abs/2611.01953"/>
  </entry>
  <entry>
    <id>https://newruntime.com/posts/aws-has-published-an-open-source-sample-for-69e9fe/</id>
    <title>AWS has published an open-source sample for migrating PHP applications to Java/Spring Boot with Kiro</title>
    <link rel="alternate" href="https://newruntime.com/posts/aws-has-published-an-open-source-sample-for-69e9fe/"/>
    <published>2026-11-06T00:00:00.000Z</published>
    <updated>2026-11-06T00:00:00.000Z</updated>
    <summary>AWS has published an open-source sample for migrating PHP applications to Java/Spring Boot with Kiro, with the same skill files reusable in AWS Transform custom transformations.</summary>
    <category term="coding-agents"/>
    <link rel="related" href="https://github.com/aws-samples/sample-php-to-java-migration"/>
  </entry>
  <entry>
    <id>https://newruntime.com/posts/deepseek-v4-1-flash-1017-03b8fa/</id>
    <title>DeepSeek V4.1 Flash 1017</title>
    <link rel="alternate" href="https://newruntime.com/posts/deepseek-v4-1-flash-1017-03b8fa/"/>
    <published>2026-11-06T00:00:00.000Z</published>
    <updated>2026-11-06T00:00:00.000Z</updated>
    <summary>Cursor’s model documentation lists DeepSeek V4.1 Flash 1017 as a selectable third-party coding model with a one-million-token context window and five reasoning levels.</summary>
    <category term="coding-agents"/>
    <category term="context-engineering"/>
    <category term="models"/>
    <link rel="related" href="https://cursor.com/docs/models/deepseek-v4-1-flash-1017"/>
  </entry>
  <entry>
    <id>https://newruntime.com/posts/hkuds-agentpilot-posts-rsi-results-9b-strategist-lifts-4be5f7/</id>
    <title>HKUDS AgentPilot posts RSI results: 9B strategist lifts DeepSeek V4 coding and Terminal-Bench scores</title>
    <link rel="alternate" href="https://newruntime.com/posts/hkuds-agentpilot-posts-rsi-results-9b-strategist-lifts-4be5f7/"/>
    <published>2026-11-06T00:00:00.000Z</published>
    <updated>2026-11-06T00:00:00.000Z</updated>
    <summary>HKUDS’s AgentPilot pairs a 9B strategy model with a larger coding model, which executes the proposed changes.</summary>
    <category term="coding-agents"/>
    <category term="evals"/>
    <link rel="related" href="https://github.com/HKUDS/AgentPilot"/>
  </entry>
  <entry>
    <id>https://newruntime.com/posts/pls-add-deepseek-v3-2-d6b02d/</id>
    <title>Pls Add DeepSeek V3.2</title>
    <link rel="alternate" href="https://newruntime.com/posts/pls-add-deepseek-v3-2-d6b02d/"/>
    <published>2026-11-06T00:00:00.000Z</published>
    <updated>2026-11-06T00:00:00.000Z</updated>
    <summary>Cursor says DeepSeek V4.1 Flash is now available as a hosted model in Cursor, selectable without a separate DeepSeek API key or a base-URL override.</summary>
    <category term="coding-agents"/>
    <category term="models"/>
    <link rel="related" href="https://forum.cursor.com/t/pls-add-deepseek-v3-2/135617/23"/>
  </entry>
  <entry>
    <id>https://newruntime.com/posts/the-old-math-is-broken-in-a-new-be39ff/</id>
    <title>The old math is broken. In a new essay, Luu argues that devs should be building bigger, more ambitious software now that agents are around.</title>
    <link rel="alternate" href="https://newruntime.com/posts/the-old-math-is-broken-in-a-new-be39ff/"/>
    <published>2026-11-06T00:00:00.000Z</published>
    <updated>2026-11-06T00:00:00.000Z</updated>
    <summary>Dan Luu argues that coding agents make it practical to attempt software projects that once seemed too expensive, from specialized developer tools to small games.</summary>
    <category term="coding-agents"/>
    <category term="developer-tools"/>
    <link rel="related" href="https://danluu.com/ambitious"/>
  </entry>
  <entry>
    <id>https://newruntime.com/posts/toptal-expertcrowd-and-hud-announce-collaboration-release-expertcrowd-0e4375/</id>
    <title>Toptal ExpertCrowd and HUD announce collaboration, release ExpertCrowd-SWE coding-agent benchmark</title>
    <link rel="alternate" href="https://newruntime.com/posts/toptal-expertcrowd-and-hud-announce-collaboration-release-expertcrowd-0e4375/"/>
    <published>2026-11-06T00:00:00.000Z</published>
    <updated>2026-11-06T00:00:00.000Z</updated>
    <summary>Toptal’s ExpertCrowd and HUD have released ExpertCrowd-SWE, a benchmark for evaluating frontier language models and coding agents on software-engineering work.</summary>
    <category term="coding-agents"/>
    <link rel="related" href="https://en.prnasia.com/releases/apac/expertcrowd-and-hud-collaborate-to-advance-ai-training-and-evaluation-release-swe-benchmark-555191.shtml"/>
  </entry>
  <entry>
    <id>https://newruntime.com/posts/baseten-launches-grid-a-gateway-and-router-for-956bed/</id>
    <title>Baseten launches Grid, a gateway and router for enterprise coding agents</title>
    <link rel="alternate" href="https://newruntime.com/posts/baseten-launches-grid-a-gateway-and-router-for-956bed/"/>
    <published>2026-11-05T00:00:00.000Z</published>
    <updated>2026-11-05T00:00:00.000Z</updated>
    <summary>Baseten opened public preview of Grid, an enterprise control layer combining a model gateway, task-level routing and central guardrails for coding agents.</summary>
    <category term="claude"/>
    <category term="claude-code"/>
    <category term="coding-agents"/>
    <link rel="related" href="https://www.baseten.co/blog/introducing-baseten-grid"/>
  </entry>
  <entry>
    <id>https://newruntime.com/signals/big-platforms-are-buying-the-tools-that-coordinate-coding-agents-and-h-97606e/</id>
    <title>Big platforms are buying the tools that coordinate coding agents, and hosted versions are shutting down</title>
    <link rel="alternate" href="https://newruntime.com/signals/big-platforms-are-buying-the-tools-that-coordinate-coding-agents-and-h-97606e/"/>
    <published>2026-11-05T00:00:00.000Z</published>
    <updated>2026-11-05T00:00:00.000Z</updated>
    <summary>Meta bought Conductor, which runs many coding agents in parallel, and Conductor Cloud shuts down on November 20.</summary>
    <category term="claude"/>
    <category term="claude-code"/>
    <category term="coding-agents"/>
    <link rel="related" href="https://www.axios.com/2026/11/04/meta-acquires-ai-coding-startup-conductor"/>
    <link rel="related" href="https://www.conductor.build/announcement"/>
    <link rel="related" href="https://www.humanlayer.dev/blog/humanlayer-joins-factory"/>
    <link rel="related" href="https://factory.com/news/humanlayer"/>
    <link rel="related" href="https://openai.com/index/openai-to-acquire-pierre"/>
  </entry>
</feed>
