Researchers found firms adopting AI coding agents saw 30% more lines of code, 20% more commits, and 23% more pull requests, but no statistically significant change in resolved Jira issues or epics. The gains got absorbed by review: the share of PRs with changes requested nearly doubled and comments per PR rose 35%.
Built on Qwen3.5 4B under Apache 2.0 and trained with reinforcement learning on about 1,500 synthetic repo tasks (no stronger model trajectories), FrogNano scores 61.5% on SWE bench Verified and 37.6% on SWE bench Pro in Microsoft's own five tool Leaf harness at 131K context. Microsoft calls it a research artifact: Python heavy, BF16 weights need about 9.3GB, and quantized builds aren't validated yet…
In the Codex desktop app, personal ChatGPT Pro users on GPT 6 Astra or GPT 6.1 Sol can now press Tab to accept a suggested follow up message, which they review before sending. It's free during the beta and doesn't use Codex limits or credits…
Taylor Wilsdon's open source Google Workspace MCP server shipped v2.1.0 on Oct 9, adding Gmail forwarding as a draft and an opt in Google Cloud Storage attachment backend for scaled out deployments. It also restores the 60 second Google API timeout that v2.0 had cut to 30 seconds…
Launched Oct 7 by developer morluto and past 10,400 stars, REA hands Claude Code, Cursor, or a terminal the ability to decompile native, Electron/JS, .NET, and web apps, trace how a feature works, and recreate it in your own stack, with v4.1 adding Android APK (JADX), firmware (Binwalk, Unblob), and IDA Pro support. Analysis runs locally, and the project says it doesn't recover original source or clone whole apps…
Out Oct 9 as a developer preview on GitHub for Linux, macOS, and Windows, LOOP works with any model behind a configurable base URL and lets you branch, pause, and resume sessions, share memory across parallel agents, and plug in MCP tools, with optional rootless container or microVM isolation. Soket's claim that it uses far less local memory than Claude Code comes from its own eight worker test and hasn't been independently checked…
Spotted something we missed? Start a thread.