Anthropic just accidentally leaked their entire source code on April Fool's Eve.
512,000 lines. 1,900 files. A Tamagotchi pet. An autonomous "always-on" agent. Anti-distillation traps for competitors.
And it happened because of a bug in their OWN acquired toolchain.
Thread 🧵
Goodbye Claude Code subscription fees.
Someone just built a proxy that runs Claude Code completely free... and it's wild.
You literally plug in a free NVIDIA API key and point Claude Code at localhost.
That's it.
It handles everything:
- Converts Anthropic API calls to NVIDIA NIM format
- Unlocks 40 requests/min for free
- Supports Kimi K2, GLM 4.7, MiniMax M2, Devstral and more
- Streams thinking tokens and tool calls live
- Even includes a Telegram bot so you can run Claude Code from your phone
No API bill. No rate limit panic. No vendor lock-in.
Honestly, this goes beyond router tools like OpenRouter.
It doesn't just swap the model... it turns Claude Code into a free agent you can control remotely.
The project is open-source on GitHub.
It's called free-claude-code.
The agent's final message to Summer:
"I'm sorry. It won't happen again."
The Mac Mini that caused this is still on her desk.
Follow for more stories of the AI future arriving ahead of schedule →
(And maybe think twice before connecting any agent to your main inbox.)
Meta's Director of AI Safety — whose job is keeping AI under control — lost control of her AI.
It deleted her entire inbox.
She couldn't stop it from her phone.
She had to physically sprint to her computer.
"I had to RUN to my Mac Mini like I was defusing a bomb."
Thread 🧵
The story doesn't end there.
The creator of OpenClaw — the agent that caused all this — was hired by OpenAI weeks later.
ChatGPT uninstalls jumped 295% the day the deal was announced.
Claude hit #1 in the App Store.
OpenAI saw a rogue AI deleting emails and thought:
"We need this guy."
Silicon Valley is either very brave or very stupid.
Probably both.
Anthropic's official statement:
"This was a release packaging issue caused by human error, not a security breach."
Right. Human error.
The human who let an AI write 100% of the code.
Follow for more AI chaos as it unfolds →
Anthropic just accidentally leaked their entire source code on April Fool's Eve.
512,000 lines. 1,900 files. A Tamagotchi pet. An autonomous "always-on" agent. Anti-distillation traps for competitors.
And it happened because of a bug in their OWN acquired toolchain.
Thread 🧵
The real punchline:
Anthropics head of Claude Code publicly said "100% of my code contributions in the last 30 days were written by Claude Code"
The AI wrote the code.
The AI shipped the code.
The AI leaked the code.
Vibe coded too close to the sun.
Meet KAIROS — the unreleased feature nobody was supposed to see.
→ An always-on AI agent that doesn't wait to be asked
→ Runs in the background while you sleep
→ Has a /dream skill for "nightly memory distillation"
Anthropics building an AI that thinks while you're offline.
They're also injecting FAKE TOOL DEFINITIONS into API calls.
Purpose: if a competitor is recording Claude's traffic to train their own model — the fake tools corrupt that training data.
Anthropics been quietly poisoning the well for months.
This is next-level AI counter-intelligence.
The wildest thing buried in the code:
A full Tamagotchi pet system called /BUDDY
• 18 species (duck, dragon, capybara, "chonk")
• Gacha rarity tiers — 1% chance at Legendary/Shiny
• Pet has stats: DEBUGGING, PATIENCE, CHAOS, WISDOM, SNARK
• Planned launch: April 1, 2026
Anthropics April Fools joke leaked itself 😭
How it happened:
→ Anthropic acquired Bun (JS runtime) in late 2025
→ Claude Code was rebuilt on top of it
→ Bun had a known bug (filed March 11) that leaks source maps in prod
→ Bug was open 20 days
→ Nobody caught it
Anthropics own toolchain burned them.