今日必看

Nvidia debuted the Open Agent Safety Platform

The failure modes of Claude Code in a guided 60h project

Show HN: Paper(Local CLI for Agent Debugging)

Hi HN, I been using LLMs for coding a lot and while they aren't awful at generating code they suck at debugging and spend endless tokens on a debugging loop. Eventually that frustrated me and I built a CLI that attempts to make AI debugging easier. How it works is any error you e

Show HN: Ferry – Sync your coding agent setup to remote Linux boxes

Claude Code Mods: plugins may now modify deeper behavior

Giving friends custom text buzzes based on Morse code

I rebuilt Claude Code's terminal UI in Rust. Then I deleted 12,000 lines of it

An agent is not an extension of the employee who created it

Blobbu – hatch your own AI agent blob (contains no AI)

Come in Alone (2026)

Worktrunk: A CLI for Git worktree management, designed for AI agent workflows

Combining cultures, from code to canvas: an interview with Ken Goldberg

Show HN: Keel – Give your agent a typed strategy language, not a trading account

I’ve been using AI agents to build systematic strategies on Hyperliquid for over a year, and slowly built up a reliable system to represent, test and deploy strategies. Typically there are a few options 1) llm in the execution loop, ie it makes decisions and trades, but that is s

QR Codes that can be read multiple different ways

Show HN: Opus6.video – Make Claude Opus 5.5 videos without a Claude account

GitHub Copilot in VS Code, September 2026 releases

This changelog covers VS Code v1.136 through v1.140 , shipped throughout September 2026. September’s releases streamline agent-driven development from implementation through pull request merge. Automations handle repeatable tasks, agent merge helps land changes, and improved sess

Actions retention now covers checks, runs, and statuses

As previously announced , checks, workflow runs, and statuses are now governed by the same GitHub Actions retention setting that controls how long artifacts and logs are kept. These records are automatically cleaned up when they exceed the retention period configured for your ent

Code coverage uploads no longer fail CI for new branches

Code coverage uploads from the GitHub Code Quality upload-code-coverage action no longer fail CI when you push a branch that doesn’t yet have an open pull request. Previously, the coverage API required a pull request number for any push to a non-default branch. Since that number

Rate limits for private vulnerability reports

Open source maintainers are receiving more low-quality and automated vulnerability reports, which can bury the reports that matter. Rate limits cap how many new reports a single account can submit in a day, both to your repository and across GitHub. This helps protect you from bu

HuGo: LLMs as Whole-Body Policy Code Designers for Humanoid Loco-Manipulation

arXiv:2609.30594v2 Announce Type: replace Abstract: For humanoids to be useful in everyday environments, they must perform a wide range of tasks that couple locomotion and manipulation. Existing approaches commonly acquire a loco-manipulation policy through reward engineering or

SimEX: Simulation-Integrated Robotics AutoResearch

arXiv:2609.38982v1 Announce Type: cross Abstract: Coding agents powered by large language models (LLMs) have shown remarkable abilities to autonomously reason about and achieve goals in the digital world. However, bringing this success to the physical world remains challenging. O

Scheduled code scanning skips inactive repositories

Weekly scheduled scans for code scanning default setup and GitHub Code Quality now start only after a push or pull request triggers an analysis, rather than counting every kind of scan as recent activity. Previously, activity for a repository was based on any unscheduled scan, in

A Proposed Rubric for Evaluating Expressed Clinical Reasoning in Large Language Model Responses

arXiv:2609.37788v2 Announce Type: replace-cross Abstract: We propose a rubric for assessing expressed clinical reasoning in model responses, drawing on three bodies of work: medical education assessment frameworks (ART, SCT, Key Feature Problems and OSCE); clinical LLM benchmarks

Agentic Tool-Augmented Reasoning for Explainable Image Forgery Detection

arXiv:2609.39066v1 Announce Type: new Abstract: Conventional image forgery detection methods produce binary scores or pixel-level masks without interpretable evidence, while recent multimodal large language model (MLLM)-based approaches generate post-hoc explanations of predeterm