AI Coding Live News

Microsoft Research, Microsoft Security and UC Berkeley open-sourced TokenEconomizer on GitHub (Apache 2.0, v0.1.0 released October 26), a plugin for Claude Code, Codex, GitHub Copilot CLI and Pi, with experimental opencode support. It learns from earlier sessions in the same repository: micro-skills remember repo facts, tool pruning drops tool definitions that go unused, and a loop guard nudges an agent that is stuck. The paper reports 18.4–30.9% lower mean cost across five agent and model combinations on 277 SWE-Bench tasks (paper figures).

Zoom

During GitHub Universe on Oct 28, Microsoft put the Microsoft Copilot harness, the shared agent execution layer behind Copilot Code, Copilot Studio, the GitHub Copilot app, the CLI and VS Code, into Foundry Agent Service as a public preview. Microsoft runs the agent loop and sandbox while developers supply instructions, skills, tools and model choice; Microsoft says it works across 20+ frontier models plus bring-your-own-key. Long-running resilience for hosted agents also reached GA. Harness runtime usage is free until Dec 1.

Zoom

On Oct 28, the day of Android Dev Summit, Google shipped Android Studio Rabbit 2 (2026.2.2) to the stable channel. The headline change is that Bring Your Own Agent (BYOA) moves from its September Canary preview to a stable feature: developers can plug any Agent Client Protocol (ACP) compatible coding agent into the IDE agent window. Google says more than 40 are available, including Anthropic Claude Agent, OpenAI Codex, JetBrains Junie and Google Antigravity, and users can sign in with personal or enterprise plans. The release also adds an agent skills manager, AI Change Review and agent-assisted LeakCanary fixes.

Zoom

Shanghai AI Laboratory has moved Atria Dawn from Preview to a full release, with open weights (plus an FP8 build) on Hugging Face under internlm/Atria-Dawn and on ModelScope, MIT licensed. Built on the 744B-parameter MoE GLM-5.2, it targets long-horizon research and engineering agent work; context doubles from the Preview's 256K to 512K and input is text only. The vendor table shows it leading the listed models on AutomationBench, CyberGym and SkillsBench, while still trailing Claude Opus 5 on coding tests such as SWE-bench Pro and Terminal-Bench 2.1 (all vendor-reported).

Zoom

On Oct 26 Anthropic released the Claude for Nonprofits plugin, a set of skills and connectors built with about 30 nonprofits. It covers the full fundraising cycle (finding funders, breaking down RFPs, drafting proposals and reports, donor stewardship), marketing, hiring and board work, and starts by having Claude map where the organization keeps its records and answer questions against the organization's own AI policy. It is installed automatically at no extra cost for verified nonprofits on Team and Enterprise plans. Anthropic says more than 25,000 nonprofits use Claude, and over 75% of them have fewer than 10 seats.

Zoom

On 2026-10-22, Anthropic's Applied AI team published "Building agents: When to fold subagents into your main agent loop" on claude.com, along with an open-source reference implementation, anthropics/single-agent-loop-kit. The article says that across several enterprise deployments, a single agent with skills frequently beat both the one-big-prompt design and the subagent design on quality, often at lower cost and latency per task; one customer that swapped its multi-agent harness for a single loop saw roughly 2x faster runtime. Anthropic also says it cut Claude Code's system prompt for more advanced models by over 80%. Subagents still make sense in three cases: measurable context bloat, parallel tasks with disposable context, or trouble selecting tools and conflicting instructions. The piece has drawn wide discussion on X over the past two days.

Zoom

On 2026-10-22 Anthropic said it is donating Agent Skills, the open standard that teaches agents a task through a folder with a SKILL.md file, to the Agentic AI Foundation (AAIF) under the Linux Foundation, the same neutral home it gave MCP. The donation covers the Agent Skills name and logo, the GitHub organization and the agentskills.io domain. Anthropic says more than 70 agent platforms, including ChatGPT, Cursor, GitHub Copilot and VS Code, have integrated the standard since December 2025, with 1 million skills published and nearly 280 million installs. Engineers from Anthropic, Google, OpenAI and others will maintain it as individuals.

Zoom

On October 21 Letta released FrontierMemory 1.0. Instead of quizzing agents on facts such as birthdays or hobbies, it gives an agent a set of past trajectories, has it write memory or skills, and scores that memory by how many coding and knowledge-work tasks a task agent then solves. Version 1.0 has 15 tasks adapted from public benchmarks such as DeepSWE, SkillsBench and APEX-Agents. On the model board (memory written in Letta Dream, tasks run in Letta Code) Opus 5.5 and GLM-5.3 both score 73.3% and Grok 4.7 scores 33.3%; on the harness board Letta's own Letta Dream ranks first. All numbers are Letta's own.

Zoom

On 2026-10-14 NEURA Robotics launched Neuron AI, a Physical AI foundation model built on a Perception-Language-Action (PLA) architecture that aligns 3D vision, touch and force, audio, robot state and language in one space, adds memory, and runs on the robot itself across five NEURA embodiments. No quantitative benchmarks were published. Customers fine-tune skills and roll them out to fleets via Neuraverse; a fully autonomous 11-minute laundry task will be demoed at NVIDIA GTC Berlin on October 20–22.

Zoom

To improve credit assignment in long-horizon agentic tasks, combining reinforcement learning with verifiable rewards (RLVR) and on-policy self-distillation (OPSD) has emerged as a leading paradigm. However, existing pipelines neglect the dynamic evolution of agentic skills and the reliability of privileged teacher signals. Researchers from CUHK and Tencent introduced RSIPO, enabling skills and policies to co-evolve recursively. RSIPO extracts skills directly from trajectory rollouts, optimizes skill generation via replay, and employs Empirical Bayes shrinkage to calibrate student trust in teacher logits. Operating on Qwen2.5-7B, RSIPO achieves 98.6% success on ALFWorld, outperforming standard GRPO by 17.4 percentage points.

Zoom

While LLM agents are rapidly deployed into sandbox workspaces with direct shell and filesystem access, post-training methods remain predominantly one-shot and bound by external teacher models. ClawLoop introduces a self-improving, harness-native training loop that operates without external teachers. Through a self-sustaining cycle of task synthesis with difficulty growth, SFT/RL training, and automated behavioral diagnosis, ClawLoop demonstrates consistent empirical gains across SkillsBench, Claw-Eval-Live, and ClawGym-Bench on Qwen3.6-27B, while successfully transferring data to 4B and 9B models.

Zoom

Reinforcement learning for language model agents is constrained by sparse episodic rewards and inefficient surface token alignment. Researchers have introduced PR-OPD (Privileged Representation On-policy Self-Distillation), discovering that injecting hindsight skills alters less than 25% of token probabilities while reconfiguring over 80% of internal hidden states. By aligning multi-layer projected representations against a stop-gradient self-teacher, PR-OPD boosts WebShop accuracy by 14.0 points and ALFWorld by 4.7 points with zero inference overhead.

Zoom