YO DOGE ARMY!! NEXT UPGRADE FOR DOGE TO MARS??
1. Insane new items + patterns
2. Bet next run for 100x leverage
3. Unlock epic Doge/rocket skins
4. Competitions w/ $DOGE jackpots
Hot take: Codex with GPT 5.5 and Claude with Opus 4.7 are still producing junior intern level code if not highly supervised and reviewed.
Even with super advanced instructions, skills and self improvement loops; i'm still amazed at the level of garbage they confidently produce.
. @AnthropicAI@claudeai first it started checking if my files were malwares at each read tool call, now it's stopping mid work... Wtf is going on those last 24h ?
So this is direct model-side injection, not in Claude Code, but in Claude models directly 🤯
Anthropic is now so scared of what their models can do, they are trying everything to make it safer
So this is direct model-side injection, not in Claude Code, but in Claude models directly 🤯
Anthropic is now so scared of what their models can do, they are trying everything to make it safer
🚨 Claude Code is now obsessed with telling me "Not malware" on every single file it reads
WTF? Every code read now comes with this mandatory disclaimer. Burning context and limits for no reason.This started in the last 12-24 hours.
Anthropic over-corrected on malware policies or what? Anyone else getting this nonstop?