AI Briefing#coding-agent
The latest AI and developer news about #coding-agent, with the original source and a short summary.
The latest AI and developer news about #coding-agent, with the original source and a short summary.
A study found that the internal representations of coding agents predict future states about 25 steps ahead of actual code modifications.
Mantis Skills, a flexible and extensible toolkit designed for security review of coding agents, has been released.
Fable 5 has a higher per-token cost than Opus 4.8, but through efficient management it lowers overall agent operating costs while improving performance.
A collection of agent skills has been released to help coding agents implement UI animations and design details with precision.
We propose FlowEval, a framework that measures whether LLM-generated UIs support natural interaction flows like real websites.
It covers technical approaches to lower and automate verification costs in order to guarantee the performance of AI agents.
Hugging Face has launched a public dataset showing usage patterns of coding agents that use the Hub.
Vercel now enables coding agents to directly view and debug execution details and traces of Agent Runs through MCP and CLI updates.
It presents ways to track and manage usage costs across various coding agents like Claude Code and Cursor.
OpenWiki is an open-source tool that automatically generates and maintains documentation so coding agents can better understand codebases.
DeepReinforce released Ornith-1.0, a coding-specialized model built on Gemma 4 and Qwen 3.5.
Research results on an evaluation framework have been released, assessing not just whether AI agents solve tasks, but how well they follow users' workflows and conventions.
You can now query web analytics data directly from the Vercel CLI using the vercel metrics command.
Vercel's AI SDK Harness adds Deep Agents and OpenCode adapters, providing unified support for various coding agent runtimes.
Instead of typing prompts directly into coding agents, 'loop engineering'—designing loops that drive the agent on its own—is becoming important.
Nvidia has unveiled ENPIRE, an autonomous research framework in which coding agents improve robot control algorithms on their own.
Chainguard has launched a security-hardened skill registry for AI coding agents.
The open-source model GLM-5.2, featuring a 1M token context and improved coding capabilities, has been released.
Xiaomi's MiMo team has unveiled MiMo Code, an open-source coding agent designed for complex, long-horizon coding tasks.
An analysis of how repository-level context files like AGENTS.md affect the performance and efficiency of coding agents
Toss's AI DX team built a 30-item Rubric system combining rule-based and model-based approaches to manage the quality of Skills for coding agents.
Anthropic stated that AI is moving toward a stage of recursive self-improvement in which AI designs and develops itself.
AI21 achieved SOTA on SWE-rebench with a 60.9% resolve rate by flipping the execution strategy to 'scale first, enrich later.'
The AI industry is expected to split into a high-margin market centered on high-performance closed models and an open-model economy.