E7Coding

In the Same Morning, OpenAI and Anthropic Both Loosened Usage Limits

Joy — Sun, 12 Jul 2026 19:00:00 +0800

On the morning of July 12, 2026, within less than one hour, two leading AI companies loosened the “usage leash.” Anthropic’s official Claude account announced at 10:02 that Claude Fable 5 availability would be extended and Claude Code’s weekly usage limit would remain 50% higher. Then OpenAI’s Tibo announced at 10:59 that the 5-hour usage window for Plus, Business, and Pro would be temporarily removed, GPT-5.6 Sol efficiency would be improved, and a usage reset was coming. Same morning, same theme: usage. This feels less like coincidence and more like an arms race heating up.

Agent Skills: Packaging Senior Engineering Discipline Into AI Coding Agents

Joy — Sat, 11 Jul 2026 10:30:00 +0800

AI coding agents have a common failure mode: they default to the shortest path. Ask one to add a feature, and it often starts coding immediately, skipping specs, tests, security review, and the practices that make software reliable. Addy Osmani’s Agent Skills aims to fix exactly that: package senior engineers’ production workflows, quality gates, and judgment into skills that agents can follow consistently, covering every stage from idea to launch.

GPT-5.6 Sol, Terra, and Luna: OpenAI Splits Agentic Coding Into Three Tiers

Joy — Fri, 10 Jul 2026 09:00:00 +0800

OpenAI previewed the GPT-5.6 family: Sol, Terra, and Luna.

The most important part is not just “another stronger model.” It is that OpenAI is explicitly splitting agentic coding models into three tiers: one frontier model, one everyday balanced model, and one fast low-cost model.

What Is A2A: How Different Agents Discover, Communicate, and Collaborate

Joy — Wed, 01 Jul 2026 15:40:00 +0800

A2A stands for Agent2Agent Protocol.

It is not about how a model calls tools. It is about how different agents discover each other, exchange messages, and collaborate on tasks.

As organizations begin deploying multiple agents internally, a practical problem quickly appears: a sales agent, support agent, finance agent, legal agent, and engineering agent may come from different teams, vendors, and frameworks. They cannot only operate inside their own systems. They also need to delegate work, pass context, and return results to one another.

What Is MCP: A Stable Way for Agents to Connect Tools and Context

Joy — Wed, 01 Jul 2026 15:30:00 +0800

MCP stands for Model Context Protocol.

Its role is clear: it provides an open protocol for connecting AI applications to external systems. Those systems may be local files, databases, search engines, business APIs, design tools, knowledge bases, or predefined prompts and workflows.

Claude Sonnet 5: A More Agentic Sonnet Model

Joy — Wed, 01 Jul 2026 09:20:00 +0800

Anthropic released Claude Sonnet 5.

The keyword for this upgrade is not “better chat.” It is more agentic: better at making plans, calling tools, using browsers and terminals, and pushing multi-step tasks forward instead of stopping halfway.

Claude Fable 5 and Mythos 5 redeployed

Joy — Wed, 01 Jul 2026 00:00:00 +0800

Update

Claude Fable 5 and Mythos 5 redeployed

July 1, 2026

Access to Claude Fable 5 and Mythos 5 is now restored.

On Friday, June 12, the US government applied export controls to our newest models, Claude Fable 5 and Claude Mythos 5. This required us to restrict access to foreign nationals, whether inside or outside the United States. Because the order took effect immediately and we had no reliable way to verify nationality in real-time, we suspended access to both models for all users.

baidu/Unlimited-OCR: A New Option for Long-Document OCR

Joy — Tue, 30 Jun 2026 13:35:00 +0800

Baidu recently open-sourced a new OCR model: baidu/Unlimited-OCR.

It is not a lightweight OCR tool for simply recognizing text in one image. It is closer to long-document parsing: converting papers, reports, books, and scanned PDFs into readable, editable text or Markdown while preserving as much structure as possible.

Five Archetypes of Future Product Teams: From Job Functions to Value Functions

Joy — Tue, 30 Jun 2026 13:00:00 +0800

I recently saw an interesting tweet: as engineering, product, design, data science, and other functions begin to merge into more hybrid roles, will future teams stop defining people by job titles and instead define them by how they create value?

Claude Skills vs. Agents: What Is the Difference and How Should You Choose?

Joy — Tue, 16 Jun 2026 16:15:00 +0800

When extending Claude, Skills and Agents are easy to confuse. Both look like ways to make Claude able to do more, but underneath they are very different. Remember the fundamental difference in one word: context. A Skill puts instructions into your current conversation. An Agent works in an isolated context and returns only the conclusion.

Helping Claude Use MCP Well: Six Prompting Tips and One Core Principle

Joy — Tue, 16 Jun 2026 15:50:00 +0800

After connecting Claude to MCP (Model Context Protocol), it can read project files, search code, and run commands. But many people find that even after setup, Claude still misses the point. The reason is simple: MCP gives Claude capability, not context. Claude must call tools before it can see your project. How you ask directly affects whether it calls the right tools and gives useful answers. This article covers six prompting tips and one core principle.

Why Multi-Agent Parallel Development Works: Git Worktree Isolation and Coordinated Merging

Joy — Tue, 16 Jun 2026 15:40:00 +0800

When you want multiple AI agents to work on the same codebase at the same time, the first wall you hit is that they share one working directory and overwrite each other’s files. Git Worktree + coordinator/worker mode exists to solve this. This article explains the principles, why it works, and where the benefits come from.

Getting Started with AI Agent Development: From Concepts to a Minimal Tool-Using Agent

Joy — Tue, 16 Jun 2026 15:00:00 +0800

AI agent development means building a system that can perceive, reason, plan, call tools, and execute tasks to achieve a goal. In one sentence: traditional AI, such as a normal chatbot, answers one question at a time; an AI agent is more like a digital worker. Give it a high-level goal, such as “analyze this company’s financial report and turn it into a 10-slide deck,” and it can break the work into steps, gather information, run code, call APIs, and deliver the result.

Understanding Recall and Recall Strategies: A Core Layer in Search, Recommendation, and RAG

Joy — Tue, 16 Jun 2026 14:00:00 +0800

In AI systems, recall and recall strategies appear in almost every scenario where you need to find things from data: search, recommendation, RAG, Q&A, risk detection, object detection, information extraction, and more. They answer one shared question: can the system find as many of the relevant things as possible?

Joining Devices to Self-Hosted Headscale: From tailscale up to headscale-ui Management

Joy — Tue, 16 Jun 2026 13:00:00 +0800

Background: the prologue explained the principles, headscale-ui added a management UI, and self-hosted DERP moved the relay closer. This final post covers the most common daily operation: how a new device logs into self-hosted Headscale with the Tailscale client and is managed in headscale-ui.

Self-Hosted DERP Relay: Stable 20ms Domestic Fallback for Headscale

Joy — Tue, 16 Jun 2026 12:00:00 +0800

Background: I self-hosted a private network with Headscale. Most devices can connect directly through WireGuard P2P, but several machines behind symmetric NAT or carrier-grade NAT always failed NAT traversal and had to use a relay. Official Tailscale DERP relays are overseas for me, often starting at 100ms+. So I built a domestic DERP relay and brought latency down to around 20ms. This post records the full process and three pitfalls.

Adding a Web UI to Headscale: headscale-ui Setup Notes

Joy — Tue, 16 Jun 2026 11:00:00 +0800

Background: in the prologue, I explained the principles of Tailscale and self-hosted Headscale. Assume you already have Headscale running. Native Headscale is CLI-only: adding nodes, approving devices, creating users, and configuring subnet routes all require SSH and headscale commands. Once you have more devices, this becomes annoying. headscale-ui is a static web frontend built by the community. It calls the Headscale API directly and brings these operations into the browser. This post records installation and two reverse-proxy / login pitfalls.

What Is Tailscale: A Practical Introduction to P2P Private Networking

Joy — Tue, 16 Jun 2026 10:00:00 +0800

This is the first post in my Tailscale lab series. Later posts will walk through a self-hosted control plane, a web UI, and a domestic relay. But before typing commands, it is worth understanding the principles. Once you know how it works, troubleshooting later becomes much easier than blindly copying commands.

Loopcraft: The Art of Stacking Loops

Joy — Tue, 16 Jun 2026 00:00:00 +0800

This is a translated and adapted excerpt from Latent Space / AINews.

There is a lot of “loop discourse” in the air:

Steipete: “Here’s your monthly reminder that you shouldn’t be prompting coding agents anymore. You should be designing loops that prompt your agents.”
Boris: “I don’t prompt Claude anymore. I write loops, the loops do the work.”
Andrej on Autoresearch: To get the most out of the tools that have become available now, you have to remove yourself as the bottleneck. You cannot be there to prompt the next thing. You need to arrange things so they are completely autonomous, maximize token throughput, and not be in the loop. The goal now is to increase your leverage. You do not want to be the researcher in the loop looking at every result and holding the system back. The question becomes: how do I refactor the abstractions so I arrange it once and hit go?

This framing is compelling, and many people do not realize how many loops we are already inside:

Anthropic Statement: US Government Requires Fable 5 and Mythos 5 to Be Disabled

Joy — Mon, 15 Jun 2026 00:00:00 +0800

Below is Anthropic’s official statement on the matter. This site republishes it as news without modification.

The US government, citing national security authorities, has issued an export control directive to suspend all access to Fable 5 and Mythos 5 by any foreign national, whether inside or outside the United States, including foreign national Anthropic employees. The net effect of this order is that we must abruptly disable Fable 5 and Mythos 5 for all our customers to ensure compliance. Access to all other Anthropic models will not be affected.

Was Claude Fable 5 Really Hacked?

Joy — Sun, 14 Jun 2026 00:00:00 +0800

Core judgment: based on public information so far, the evidence better supports “safety guardrails were bypassed / model jailbreak succeeded” rather than “Anthropic servers were breached, model weights were stolen, or user data leaked.” The value of this incident is not panic. It is a reminder that as frontier models become more capable, safety boundaries must move from single-model guardrails to system-level governance.