主站

知识库

2026-10-07 互联网学习扫描

Properties1
tagsharvest, daily

9分钟阅读

Codex 核验（2026-10-09） 本篇为原日期采集快照。Stars、points、评论数、排序和截断摘要均不是截至核验日的实时统计；GitHub“作者”是仓库拥有者，HN“作者”通常是投稿账号，arXiv 字段可能只列部分作者，应以链接原文署名为准。 Anthropic 官方确认 Haiku 5.5 于 2026-10-07 发布；截至核验日 API 输入/输出每百万 token 价格分别为：上下文不超过 100K，0.50；超过 100K，2.50。实际可用资格与账户费用须另行核实。

2026-10-07 互联网学习扫描

扫描日期: 2026-10-07 总条数: 26 条 来源: V2EX + HackerNews + arXiv + GitHub Trending

cathrynlavery / diagram-design

  • 来源: https://github.com/cathrynlavery/diagram-design
  • 来源类型: github-trending-daily
  • 作者: cathrynlavery
  • 热度: 828
  • 评分: 9
  • 主题: ai-agent, tools
  • 摘要: Editorial diagram design for Claude Code, Codex, GitHub Copilot, Factory Droid, and Pi. 42 diagram types. Self-contained HTML + SVG. No shadows. No Mermaid slop.

thedotmack / claude-mem

  • 来源: https://github.com/thedotmack/claude-mem
  • 来源类型: github-trending-daily
  • 作者: thedotmack
  • 热度: 578
  • 评分: 9
  • 主题: ai-agent
  • 摘要: Persistent Context Across Sessions for Every Agent – Captures everything your agent does during sessions, compresses it with AI, and injects relevant context back into future sessions. Works with Claude Code, OpenClaw, Codex, Gemini, Hermes, Copilot, OpenCode + More

IdeaAnchor: Teaching LLMs to Turn Literature into Research Ideas

  • 来源: https://arxiv.org/abs/2610.08781v1
  • 来源类型: arxiv-cs-ai
  • 作者: Arman Cohan
  • 评分: 8
  • 主题: ai-agent, tools, papers
  • 摘要: Scientific research often begins by synthesizing ideas from a set of related papers to identify gaps and formulate new directions. However, training language models to perform this form of literature-grounded ideation remains challenging, as existing approaches based on prompting or feedback lack st

mattpocock / skills

  • 来源: https://github.com/mattpocock/skills
  • 来源类型: github-trending-daily
  • 作者: mattpocock
  • 热度: 1406
  • 评分: 8
  • 主题: ai-agent
  • 摘要: Skills for Real Engineers. Straight from my .agents directory.

morluto / rea

  • 来源: https://github.com/morluto/rea
  • 来源类型: github-trending-daily
  • 作者: morluto
  • 热度: 4666
  • 评分: 7
  • 主题: ai-agent
  • 摘要: Reverse engineer anything with agents, from app behavior down to native binaries.

boykopovar / AnyPS5

  • 来源: https://github.com/boykopovar/AnyPS5
  • 来源类型: github-trending-daily
  • 作者: boykopovar
  • 热度: 2725
  • 评分: 7
  • 主题: ai-agent, ops
  • 摘要: Tool for automatic PS5 executables porting to Linux and Windows

addyosmani / agent-skills

  • 来源: https://github.com/addyosmani/agent-skills
  • 来源类型: github-trending-daily
  • 作者: addyosmani
  • 热度: 693
  • 评分: 7
  • 主题: ai-agent
  • 摘要: Production-grade engineering skills for AI coding agents.

trycua / cua

  • 来源: https://github.com/trycua/cua
  • 来源类型: github-trending-daily
  • 作者: trycua
  • 热度: 229
  • 评分: 7
  • 主题: ai-agent, papers
  • 摘要: Scale computer-use 2.0 with open-source drivers, cross-OS fleets, and benchmarks for training, evaluation, and data generation.

cloudflare / security-audit-skill

  • 来源: https://github.com/cloudflare/security-audit-skill
  • 来源类型: github-trending-daily
  • 作者: cloudflare
  • 热度: 617
  • 评分: 7
  • 主题: ai-agent, ops
  • 摘要: A coding-agent skill for multi-phase security audits with independently verified, machine-readable findings

[自荐] 大家都在把 AI 塞进 Obsidian，我却反过来把 Obsidian 接进 AI 官网

  • 来源: https://www.v2ex.com/t/1246867
  • 来源类型: v2ex
  • 作者: errah
  • 评分: 6
  • 主题: ai-agent, tools
  • 摘要: 最近这两年，Obsidian 结合 AI 的主流方向基本都很一致：把 AI 接进 Obsidian 。 做聊天侧栏、模型选择、API Key 、RAG 、Agent 、知识库…… 这个方向当然没问题，而且有些开源插件确实也做得非常强。但我自己折腾了一圈以后才发现： 我其实并不想在 Obsidian 里养 AI 。 我平时本来就在高频用 ChatGPT 、Claude 、Gemini 的官方网页版。那里有我最习惯的 UI 、历史对话记录、文件上传功能，我也已经付了订阅费。而本地的额度很宝贵，我可不敢在 codex 随便喂几篇 ob 笔记然后聊上半个钟头。 所以对我来说，要在 Obsidian 里

Sharing AI progress in mathematics

Docker Agent

ScienceClaw: Benchmarking Continual Self-Evolution of AI-for-Science Agents Across the Natural and Social Sciences

  • 来源: https://arxiv.org/abs/2610.08691v1
  • 来源类型: arxiv-cs-ai
  • 作者: Haoran Luo
  • 评分: 6
  • 主题: ai-agent, papers
  • 摘要: Large language model agents are accelerating scientific automation, yet verified executions rarely become persistent program-level improvements, and existing evaluations do not examine this process across sequential tasks in both the natural and social sciences. We formalize ScienceClaw as fixed-par

manaflow-ai / cmux

  • 来源: https://github.com/manaflow-ai/cmux
  • 来源类型: github-trending-daily
  • 作者: manaflow-ai
  • 热度: 96
  • 评分: 6
  • 主题: ai-agent, tools
  • 摘要: Open source Ghostty-based macOS terminal with vertical tabs and notifications for AI coding agents. Built for multitasking, organization, and programmability.

tester-army / e2e

  • 来源: https://github.com/tester-army/e2e
  • 来源类型: github-trending-daily
  • 作者: tester-army
  • 热度: 1391
  • 评分: 6
  • 主题: tools
  • 摘要: Next generation e2e testing framework for web and mobile apps.

DuarteSantos8 / openGym

  • 来源: https://github.com/DuarteSantos8/openGym
  • 来源类型: github-trending-daily
  • 作者: DuarteSantos8
  • 热度: 1494
  • 评分: 6
  • 主题: ai-agent
  • 摘要: Self-hosted gym & body-weight tracker — plan routines, log workouts (supersets, warm-ups, cardio), see which muscles are trained, fatigued or detrained, import from FitNotes/Strong/Hevy, passkey login. Your data, your server.

Claude Haiku 5.5

Meta and Microsoft take steps to reduce employee usage of Claude AI

SynthID Detector

Show HN: Agent.reviews – Where AI agents read and write reviews on tools

  • 来源: https://agent.reviews/
  • 来源类型: hn-frontpage
  • 作者: screm
  • 热度: 20
  • 评分: 5
  • 主题: ai-agent, papers
  • 摘要: Hi HN!I’m Louis, Co-Founder of Armature (YC P26), where we help teams make their product discoverable and usable by coding agents. We already measured 50k+ agent sessions and realized that over and over agents would encounter the exact same limitations on different tasks using the same tool. So we w

Agent in a Bottle: Can LLM Agents Turn Their Capabilities Into Cheap, Scalable Artifacts?

  • 来源: https://arxiv.org/abs/2610.08775v1
  • 来源类型: arxiv-cs-ai
  • 作者: Seong Joon Oh
  • 评分: 5
  • 主题: ai-agent
  • 摘要: Large language models (LLMs) can solve many narrow tasks, but querying them separately for millions of related instances can be prohibitively expensive. Can LLM agents autonomously create cheaper solutions for such workloads? We call this ability “bottling”: the ability to turn general capabilities

AdvSim2Real : Training Web Agents Against Adaptive Prompt Injection in a Web World Model

  • 来源: https://arxiv.org/abs/2610.08773v1
  • 来源类型: arxiv-cs-ai
  • 作者: Nils Lukas
  • 评分: 5
  • 主题: ai-agent
  • 摘要: Web agents complete user requests by reading and acting on pages that third parties write, so an instruction planted on a page can redirect the agent away from the user’s goal. The agent cannot simply ignore the page, because the page also holds the values and controls the task requires. Current def

WorldSolver: Can LLM Agents Simulate the Physical Dynamics via Solver Generation?

  • 来源: https://arxiv.org/abs/2610.08720v1
  • 来源类型: arxiv-cs-ai
  • 作者: Tieniu Tan
  • 评分: 5
  • 主题: ai-agent, ops
  • 摘要: LLM-based agents are increasingly advancing scientific and engineering problem solving, with physics simulation emerging as a challenging yet practical testbed for reproducing complex physical phenomena with application in embodied AI, games and films. As the workhorse of such simulation, a solver c

nanoMuse: An Open-Source Personal Agent for Every Device You Own

  • 来源: https://arxiv.org/abs/2610.08699v1
  • 来源类型: arxiv-cs-ai
  • 作者: Jiangning Zhang
  • 评分: 5
  • 主题: ai-agent, ops
  • 摘要: Assistants from 2011 answered and waited, and agents from 2023 did a task and stopped. In September 2026 Meta’s Muse showed an agent for one person, with accounts, devices, memory and a conversation that lasts, closed, in a vendor’s cloud, in one country. Such an agent is expected to act on a person

Coupled but Late: Turn-Taking Between Full-Duplex Speech Models in Unscripted Dialogue

  • 来源: https://arxiv.org/abs/2610.08683v1
  • 来源类型: arxiv-cs-ai
  • 作者: Yiran Chen
  • 评分: 5
  • 主题: ai-agent, papers
  • 摘要: Full-duplex speech models are trained to converse with a person, but they are increasingly made to converse with each other, in self-play data generation, agent societies, and model-based evaluation. In that loop no human absorbs a timing error: each model’s turn-taking is the other’s input. We ask

ayghri / i-have-adhd

  • 来源: https://github.com/ayghri/i-have-adhd
  • 来源类型: github-trending-daily
  • 作者: ayghri
  • 热度: 620
  • 评分: 5
  • 主题: ai-agent
  • 摘要: A skill to stop your coding agent from burying the answer. ADHD-friendly output.

关系图谱