Aryko

[花叔 @AlchainHust]

https://t.co/QXMZggJmk4

[Aryko @Aron4821]

@aqqqq88 @telegram 可以帮忙注册，私我就行，一天内回复

[0x_Miko @Mikocrypto11]

Anthropic 自己的团队，刚刚演示了如何搭建 production AI agents

37 分钟

免费
而且是由真正构建这套系统的 engineers 亲自讲
这不是又一个“AI 很强”的泛泛分享

而是直接展示：
怎么把 AI agents 从 demo，推进到 production
很多人过去 6 个月，还在手动管理每一个 workflow
自己拆任务
自己复制上下文
自己盯执行
自己检查结果

但 Anthropic 这次讲的是另一件事：
如何把这些 workflow 放到 autopilot 上
让 agent 不只是回答问题，而是参与执行、协作和持续推进任务

真正有意思的是：
AI agent 的关键，不是单次生成得多漂亮
而是能不能稳定接住真实工作流
能不能连接 tools
能不能处理上下文
能不能持续执行
能不能进入 production

先看这个 37 分钟 workshop
未来真正拉开差距的，是谁会用 AI 提问，还是谁能把 AI agents 接进自己的生产系统？

[Suryansh Tiwari @Suryanshti777]

Developers are quietly building a second brain inside their AI coding tools right now.

And almost nobody realizes how important this shift is.

Claude Code’s "Skills.md" looks like “just another markdown file.”

It’s not.

It’s the beginning of programmable AI memory for software engineering.

Most AI coding workflows today are broken in the same way:

• repetitive prompting
• inconsistent outputs
• forgotten context
• unstable architecture decisions
• AI behaving differently every session

So developers waste hours re-explaining:
“use this stack”
“follow this pattern”
“don’t break this API”
“write code this way”

Over and over again.

"Skills.md" changes that completely.

Instead of prompting the AI every time…

You teach it how your team builds once.

Now Claude starts operating with:

- coding standards
- architecture rules
- debugging workflows
- testing systems
- project conventions
- review patterns
- deployment logic

persisted directly into its workflow.

That’s a massive shift.

Because this is where AI stops feeling like autocomplete…

and starts feeling like infrastructure.

The real breakthrough in AI coding isn’t smarter models anymore.

It’s persistent operational context.

The teams moving fastest right now are not the ones writing better prompts.

They’re the ones building reusable intelligence layers around AI.

That compounds.

Fast.

And the scary part?

Most developers still haven’t noticed this transition happening.

They’re benchmarking models.

Meanwhile advanced teams are quietly building internal AI operating systems around their workflows.

That gap is going to become enormous.

"Skills.md" is interesting because it represents something much bigger than Claude Code itself:

AI that adapts to YOUR engineering system instead of forcing engineers to adapt to the AI.

That changes:
• onboarding
• consistency
• code quality
• iteration speed
• institutional memory
• engineering velocity

A lot of people will look back at these “simple markdown skill files” the same way we now look at:

- Dockerfiles
- package.json
- .gitignore
- tsconfig

Small files.

Huge workflow shift.

The future of AI coding won’t belong to the people with the longest prompts.

It’ll belong to the people who build the best AI environments.

And that transition has already started.

[Y11 @seclink]

冷知识，三个最核心的信息差点：

1. OpenCode — GitHub Stars 已超越 Claude Code（160K+ vs 122K+），中国几乎无人讨论
2. Gemini CLI — 1000 请求/天免费，对中国成本敏感用户极具吸引力，无报道
3. Goose/OpenHands — 代表"自主编码 Agent"方向，中国认知几乎为零

[Aryko @Aron4821]

@TooooooBug 9929 线路的非常稳定、延迟低，缺点是带宽比较小，50Mbps 就意味着你下载东西的最快速度是大概 6.28MB /秒.
4837 线路就是 163 网，晚高峰延迟会变高，丢包率也会变高，不过带宽大（下载东西的速度快，但高峰期可能速度还不如 9929 线路）

[Aryko @Aron4821]

It's suitable for context creator.

[Geek @geekbb]

有点意思，Codex 用的 Skill，专门给文章配手绘风格的插图，把文章里抽象的想法变成看得懂的手绘配图。

https://t.co/4a47xdShxz

[九原客 @9hills]

gpt-5.5 就是无情的干活机器，plan、design、review、test等都用deepseek-v4-pro

每晚一个 Goal，拆成Phase后，每个Phase先写测试、再写代码、Code Review、Simplify、运行测试、修正、Commit。

[Nico @nicos_ai]

Mejores cuentas de cada lab de IA para mantenerte siempre informado:

Anthropic
→ @karpathy → cuenta imprescindible en IA, acaba de unirse a Anthropic
→ @bcherny → creador de Claude Code, siempre comparte tips muy útiles
→ @trq212 → también desarrollador de Claude Code, escribe artículos increíbles sobre CC

OpenAI
→ @polynoamial → trabaja en investigación de razonamiento, comparte muchos detalles técnicos
→ @gabriel1 → desarrollador de Sora, trayectoria profesional muy interesante
→ @jxnlco → enfocado en experiencia de desarrollador, comparte mucho sobre Codex

Google AI
→ @OfficialLoganK → todas las actualizaciones importantes de Gemini y AI Studio
→ @ammaar → producto y diseño, comparte cosas geniales sobre vibe-coding en AI Studio
→ @fofrAI → casos de uso muy creativos con modelos generativos

Cursor
→ @leerob → la voz más activa detrás de las novedades de Cursor
→ @ericzakariasson → comparte muy buenos insights sobre cómo usar Cursor
→ @mntruell → CEO de Cursor, publica los lanzamientos y datos de uso más importantes

xAI
→ @milichab → se unió hace poco a xAI, comparte novedades sobre Grok
→ @skcd42 → también cubre los lanzamientos grandes de Grok
→ @elonmusk → Elon hace muy buen trabajo reposteando y dando visibilidad a todos los productos de xAI

Me he dejado a alguien?

[Rahul @sairahul1]

Anthropic just officially released the blueprint for creating a company with Claude Code and it's mind-blowing😭

CEO: 1 human (who sleeps)
Employees: several AIs
Activities: the AIs divide up the tasks and move forward on their own

Work is literally dying... I've summarized the full guide below, read it when you've got 5 min ⤵️

If you want the AI to work while you sleep → save this as a bookmark 🔖

[Aryko @Aron4821]

@JieTwo26168 @xingjia520 搬瓦工的 VPS 不也很贵吗

[ClaudeDevs @ClaudeDevs]

New in Claude Code (research preview): dynamic workflows.

Claude writes an orchestration script on the fly, then spins up a large fleet of coordinated subagents in parallel to take on your most complex tasks.

Use the word "workflow" in a prompt to get started.

[Avid @Av1dlive]

Anthropic's own team just showed how to build production AI agents.

37 minutes. free. from the engineers who built it.

watch the workshop. bookmark it.

you spent 6 months managing every workflow yourself.

they just showed how to put all of it on autopilot.

Then read the guide below.

[cat @_catwu]

Excited to share our most powerful new Claude Code feature: dynamic workflows!

Mention "workflow" in a prompt and Claude will dynamically create an orchestration plan that it strictly follows, allowing you to confidently trust that every stage happens in the right order even across 100s of agents.

[Aryko @Aron4821]

@shxingwng213670 @crabcrcu 是充了 Plus 就不用验证码验证了？

[Nainsi Dwivedi @NainsiDwiv50980]

Andrej Karpathy just explained the future of software engineering without directly saying it.

The best AI engineers are no longer “prompting.”

They’re building systems around the agents.

Karpathy’s biggest insight wasn’t:
“Claude can code.”

It was:
LLMs become dramatically better when you force them into disciplined workflows.

That’s why "CLAUDE.md" files are suddenly everywhere.

Not because they’re prompts.

Because they behave like an operating system for the agent.

Karpathy called out the exact problems with AI coding:

- models assume instead of asking
- they overengineer simple tasks
- they hide confusion
- they rewrite unrelated code
- they optimize for completion, not correctness

So developers started encoding rules directly into the workflow:

→ Think before coding
→ Simplicity first
→ Surgical edits only
→ Goal-driven execution

And the results are wild.

People are now running multiple Claude Code agents in parallel like engineering teams:

• one agent researching
• one debugging
• one writing tests
• one optimizing code
• one validating outputs

Not “AI assistance.”

Actual orchestration.

And this part from Karpathy changes everything:

“Don’t tell the model what to do. Give it success criteria and let it loop.”

That is the shift.

From:
“write this function”

To:
“here’s the goal, constraints, tests, and verification system — now iterate until correct.”

The craziest part?

This already feels like a phase shift in engineering.

A lot of developers quietly went from:
80% manual coding → to 80% agent-driven coding in just months.

Not because AI became perfect.

Because the leverage became impossible to ignore.

We’re entering an era where the highest leverage engineers won’t necessarily be the best coders.

They’ll be the people who build the best systems around AI agents.

[Akshay 🚀 @akshay_pachaar]

Microsoft just open-sourced SkillOpt!

A framework for training agent skills like neural networks:

SkillOpt treats a plain markdown file as the trainable parameter of a frozen LLM agent, applying the same optimization discipline used in weight training: learning rates, validation gates, batch sizes, and epoch schedules.

The analogy maps precisely. The skill document is the parameter. Trajectory-derived edits are the gradient direction. An edit budget is the learning rate. A held-out split is the validation check.

Here's how it works.

A frozen model runs tasks with the current skill and produces scored trajectories. A separate optimizer model analyzes failures in minibatches, proposes structured add/delete/replace edits, and ranks them under a budget cap.

If the candidate skill improves performance on a held-out split, the edit is accepted. If not, it's rejected and stored so the optimizer avoids repeating failed changes.

The deployed output is a single best_skill. md file, typically 300 to 2,000 tokens. No weight changes, no extra inference-time calls.

The learned rules are compact and readable. These read like rules a thoughtful engineer would write after a day with the benchmark, except they were discovered automatically.

Learn more:

Paper: https://t.co/sdj5DW7t9h
GitHub: https://t.co/W3DcpBCni0

SkillOpt isn't the first system to treat skills as something you can optimize.

Hermes Agent independently built the same idea through a combination of skill_manage, Curator, and an optimization loop called GEPA that scores, mutates, and promotes skill documents across runs.

Two teams, different architectures, same conclusion: the skill file is the highest-leverage thing to optimize in a frozen-model agent.

I wrote a deep dive on how the Hermes agent works and covered all of these topics briefly.

The article is quoted below.

[GitHub Projects Community @GithubProjects]

CodeGraph replaces file-grepping with a pre-indexed knowledge graph so coding agents spend less time and money exploring your codebase.

- 70% fewer tool calls by querying symbol relationships instead of scanning files
- 35% cheaper on average across 7 real-world codebases
- Works with Claude Code, Cursor, Codex, opencode, and Hermes Agent
- Self-contained binary, no Node.js required for installation

[Shubham Saboo @Saboo_Shubham_]

This is ACTUALLY context engineering for your AI coding agents.

It turns any codebase into an interactive graph your agent can query.

Works with Claude Code, Codex, Antigravity.

100% Opensource.

[Sac @Saccc_c]

99% 的人使用 Codex 中的/goal 命令都是错误的

大多数的人使用方式： /goal，帮我完成xx任务

正确的使用方式如下图（<>是你需要撰写的文字）：