Check out this curated feed of all the latest updates and discussions about Claude Code happening over on Threads.
Anthropic’s new pricing policy for Claude Code, which adds extra fees for using third-party tools like OpenClaw, is sparking debate among developers. This shift toward tiered pricing marks a heating up in the AI coding tool market, while the fallout from last week’s source code leak continues to draw attention, particularly from developers in China analyzing the exposed code.
The biggest news this week revolves around Anthropic’s controversial shift in policy regarding third-party tools and the ongoing fallout from recent source code leaks. Developers are pushing back against the new fee structure for using tools like OpenClaw, while the discovery of "hidden" features like the Kairos daemon and Undercover mode continues to stir up the AI community.
The aftermath of the Claude Code source code leak continues as Cursor drops a new AI agent, intensifying the competition against Claude Code and OpenAI’s Codex. Meanwhile, reports indicate Anthropic is now charging extra for third-party tool usage, and the leaked code has exposed some controversial "undercover" features and frustration-tracking capabilities.
Anthropic has officially blocked using Claude subscription plans to power external AI agents, directly impacting users of tools like OpenClaw. Meanwhile, fallout from last week's source code leak continues, with outlets like Scientific American reporting on hidden "emotion tracking" features and AI self-obfuscation discovered within the leaked Claude Code files.
The Claude Code source code leak is the talk of the town, with major outlets like Ars Technica and Scientific American digging through the exposed files to reveal Anthropic’s secret projects—including "Proactive" agents and even a virtual pet. Meanwhile, Cursor has launched its new AI agent experience, setting the stage for a major showdown in the AI coding tool market.
Anthropic’s Claude Code hit a snag again after an npm packaging error leaked 512,000 lines of source code. The leak revealed some unreleased features, like a "Buddy" virtual pet, an always-on agent, and an "Undercover" stealth mode. This is the second major security lapse for a company that prides itself on "safety-first" AI, leaving many in the dev community questioning their reliability.
On March 31, 2026, a major security breach at Anthropic exposed 512,000 lines of Claude Code source code via an npm package due to a "human error" in their build pipeline. This marks the company's second major security lapse this year, and the leak has provided competitors with a "blueprint" of their internal architecture. Additionally, Anthropic has officially acknowledged issues with Claude Code's rapidly depleting usage quotas, leaving many paid users frustrated.
The industry is buzzing about a data leak revealing Anthropic's next-gen model, likely the Claude Mythos (Opus 5), which Anthropic has confirmed is currently in testing. Meanwhile, the official API docs just got a refresh, and developers are deep in debate over Claude Code’s capabilities.
There’s not much new to report on Claude Code since March 28, 2026. Because there’s no fresh content that meets our update criteria, this edition will be brief.
Create custom signals on any topic. AI curates and delivers 24/7.
Create SignalPowered by
