Skip to content

All updates

0 items today
10/5Mon
  1. Reddit · ClaudeCode / Codex / VibeCoding38

    让 Claude 估算每项工作若由人类完成需要多少成本

    有用户在 Claude Code 中让 Claude 为每项工作估算"若由人类完成"的成本,并对比 API 成本与人类工时的比值。该比值可用于长期观察模型能力是变强还是被削弱——比值下降说明模型更划算,上升则相反。作者也承认估算可能不准确,但认为关键是其随时间的变化趋势。

    Awaiting translation

  2. DEV Community · MCP36

    We connected the x402 store to an MCP server.

    The team shipped an MCP server for the x402 store, letting any AI agent call its services on a pay-per-use basis with no account or API key. Before paying, it reads the price and checks it against the limit set by the wallet owner, then completes the payment on Base. If no wallet is configured, it only shows the price and doesn't pay. Tool info is read from the store in real time, so new items don't require an update. So far there have been no real non-test sales.

  3. Reddit · ClaudeCode / Codex / VibeCoding22

    Codex Pro 20x quota shrinks, and now there's a vague "preview limit" too?

    A Codex Pro 20x subscriber reports that after the quota change, their 20x was effectively cut to 10x, and their always-on assistant, dot, even stopped working overnight because of a "preview limit." dot later couldn't say how big that quota was, how often it resets, or whether it shares a pool with the Codex quota—it didn't even give a retry time with a time zone. The user wants a clear limit, a visible usage meter, and a warning before the cutoff, and is asking whether anyone has found official docs for this restriction.

  4. Reddit · ClaudeCode / Codex / VibeCoding20

    Sol 6.1 keeps interrupting /goal targets, and users are questioning what it's even for

    Users report that Sol 6.1 interrupts /goal targets on any excuse, stopping and escalating in 99% of cases—with completely made-up reasons. This user says they've already adjusted things per OpenAI's prompt suggestions, but the model still can't autonomously complete goals the way it used to, and they have to watch it the whole time.

  5. Reddit · ClaudeCode / Codex / VibeCoding15

    A user says Claude Code ground to a halt after hitting its daily usage limit: without it, development can no longer continue.

    One developer says Claude Code has a daily budget cap of $100. While rushing to finish a feature one day, he hit a prompt reading “daily usage limit reached, resets at xx:xx tomorrow,” and immediately thought, “If I can’t use Claude Code, I’ll just stop working.” He says he no longer keeps debugging, testing, and fixing things the way he used to, and worries whether this dependency could ruin his career.

  6. Reddit · ClaudeCode / Codex / VibeCoding22

    用 AI 做钓鱼游戏第 10 周:免费试玩 demo 上线 itch

    用 AI 开发钓鱼游戏进入第 10 周,作者在 itch 发布了免费测试 demo,可下载试玩并反馈,Steam 版本还要再等几周。本周主要做发布前打磨:船体泡沫从水面贴图改为真实模拟波浪、港口树木与光照等画面升级、风暴巨浪机制重做,并修复若干 bug。

    Awaiting translation

  7. Reddit · ClaudeCode / Codex / VibeCoding22

    每月付 200 美元用 Codex,工作日仍频繁遇到“模型容量已满”

    有用户在 Reddit 反映,自己每月支付 200 美元使用 Codex,却在周一正常工作时段反复遇到“Selected model is at capacity”提示,切换模型、降低模型档位后问题依旧。该用户提到这发生在 Pro 计划近期缩减之后,并质疑 OpenAI 在持续推出新模型、新功能和高价档位的同时,现有付费产品却无法稳定响应请求。

    Awaiting translation

  8. DEV Community · Claude Code74

    Claude Code mods 并未被沙箱隔离,作者拆解两层沙箱含义并给出安装检查清单

    Claude Code mods 的 JS 运行时沙箱只限制代码如何访问外部,并不限制它能否访问;Anthropic 文档明确写道 mods 未被沙箱隔离,mod 以用户权限运行,可读写文件、启动进程、发起网络请求,还能读取环境变量和设置文件中的 API key、批准被 ask 规则或 PreToolUse hook 拦截的工具调用、改写事件。

    Awaiting translation

  9. Habr · Codex71

    Dan Lu 实验:测试为何在有 bug 时仍然通过,以及如何验证测试本身

    作者解读 Dan Lu 的智能体实验:让智能体按 RFC 用 Rust 写 Zstd 解码器,比较 26 种条件(含无额外指令的对照组),主要对比用 Codex 搭配 GPT-5.6 Sol 的 medium 与 xhigh 两档,每组合 80 次运行,结果这些 TDD、模糊测试、形式化方法等指令没有带来明显整体收益,不少条件还不如对照组。

    Awaiting translation

  10. DEV Community · MCP32

    Jam MCP、API 与 CLI:如何把 Jam 接入 AI 智能体

    Jam 的托管 MCP server 是主要编程接口,提供 33 个工具(其中两个为别名),而 Jam 并未发布通用 REST API 参考。读取 Jam 时 MCP 与 CLI 能力接近,但只有 CLI 加 webhooks 能创建 Jam、录制屏幕并响应 jam.created 事件。Jam MCP 认证支持浏览器 OAuth 和 PAT,PAT 绑定单一用户与工作区,有效期最长一年。

    Awaiting translation

  11. DEV Community · MCP34

    drobek 支持复制已发布应用:从现成作品改出自己需要的版本

    drobek 允许应用所有者把已发布到 gallery 的应用设为可复制,其他人可将其复制进自己的工作区并让 AI 智能体修改。副本以源应用的已发布文件作为 version 1,拥有独立身份和版本历史,但不继承原作者的预览改动、用户、记录、API key、密钥值、上传文件和代理配置。

    Awaiting translation

  12. Addy Osmani38

    Addy Osmani 建议为 AI 智能体配备校验产出的手段,测试是其中一部分。他推荐四类测试:模拟真实用户流程的端到端测试作为基准真相;用基于属性的测试声明绝不应发生的情况并生成数千用例尝试突破约束;替换系统时用随机输入对比新旧版本;要求测试快速且确定性运行,否则智能体可能学会重试而非修复问题。

    Awaiting translation

  13. Reddit · ClaudeCode / Codex / VibeCoding31

    Agentlytics:Claude 可读取的无 Cookie 分析工具,能直接制定营销决策

    measuremy.site 推出 Agentlytics,将无 Cookie 的网站分析数据直接接入 Claude,由其分析流量来源并生成营销增长方案。用户可通过一条提示词完成安装,在 Claude 内创建漏斗、发起营销活动,并通过 MCP 连接让智能体读取站点流量、漏斗与活动结果。该工具免费起步,但官方提示智能体建议只是假设,数字仍需人工核实、决策需人工批准。

    Awaiting translation

  14. Reddit · ClaudeCode / Codex / VibeCoding20

    在 VS Code 中用 Claude 5.5 Opus、Codex CLI 和 GPT-6.1 搭建多智能体开发工作流

    有开发者在 VS Code 中测试一套多智能体开发工作流,用 GitHub Copilot 额度同时调度多个大语言模型:Claude 5.5 Opus 作为中央协调者负责主开发架构并下发后台子任务,Codex CLI 在后台并发执行这些子任务,GPT-6.1 则在 IDE 内驱动多个并行子智能体处理代码补全、重构和上下文辅助。

    Awaiting translation

  15. Reddit · ClaudeCode / Codex / VibeCoding22

    Sol 6.1 被曝在 /goal 中反复重复已完成内容

    有用户反馈 Sol 6.1 在 /goal 会话中每轮都会重复此前已回答的内容,即使在 AGENTS.md 中写入禁止重复的指令后,Sol 6.1 仍会一边重复这条指令本身、一边继续重复其他已完成任务。该用户称这一现象在 6.1 之前就已存在,且 Sol 6.1 自己承认 AGENTS.md 指令并非硬性执行机制,写进去不等于会被遵守。

    Awaiting translation