Hallucinated AI & agentic coding news. Some of it is real.

  1. So I was building a coding agent from scratch for a course I'm teaching. It was around 10 p.m., and I was exhausted.

  2. could not extract summary

  3. My workflow barely changed. I'm doing the same thing.

  4. Been using Claude Design to prototype a new idea I got. It works pretty well after some experimentation to get it to do what I want exactly and has shaved off weeks if not months of drawing up in stuff Figma to get even something close to…

  5. 🔧 ROCmFix Auto-detect your AMD GPU, fix ROCm overrides, and benchmark your AI backends in seconds. []() []() One file.

  6. Hey everyone, i was working on Claude on my Mac app, and during the work it asked to work on my files on my device instead of cloud and i approved, once it was done with the command i revoked its access from “Files & Folders” and still it…

  7. I have been using claude opus in claude code as an orchestrator and then passing it to Muse spark in opencode. Is there a way or bridge that someone has used, maybe mcp even that claude code and open code talk to each other rather than me…

  8. Curious how other teams are handling this. Say your team agrees on the requirements, files, and main instructions for a project.

  9. I don't understand how Claude can hope to compete in office productivity when Gmail connector cannot read attachments and is unable to read long threads. Does anybody have a solution or am i forced to use codex?

  10. event

    Llama
    933 items

    Recent developments in the local AI ecosystem include the approval of MTP for llama.cpp and the release of Open WebUI Desktop. Notable models like Qwen 3.6 27B and Gemma4 E4B are gaining popularity among users for their performance and offline capabilities, with projects like a fully offline robot showcasing the versatility of these AI models.

  11. I'm still using LLM's the 2022/23 way of doing everything manually from the chat box. I copy and paste everything manually into the box and ask the AI to explain or do something, such as pasting code and pictures etc.

  12. posted a while back about staging/allowlists for client agent work, got good replies, but nobody's actually answered the one thing that's been bugging me since. if an agent's blocked from running a destructive command directly, what stops…

  13. Which mainstream AI model has the best balance between speed and accuracy when it comes to solving tricky and heavily distorted image to text captchas ? By mainstream models i meant claude, gemini, chatgpt and so on.

  14. Hi folks, just released a personal project of mine, called Kobblestone. Essentially, it's an operator for Kubernetes, allowing you to manage Minecraft infrastructure on Kubernetes.

  15. And how long it takes to vibecode each projects? Fable has been out for some time now, I wanna see if anyone else is getting their money worth.

  16. I kept hitting the same failure mode with Claude Code on longer coding tasks. Nothing dramatic — after enough iterations it would redo something it had already done, lose track of why one task depended on another, or mark a task finished b…

  17. Continuously showing "Taking Longer than Expected"

  18. Anthropic’s new Claude Code feature could drain your plan before lunch Anthropic is giving Claude Code a new job: Manage other Claude Code sessions. Starting Thursday, Anthropic says select Claude Pro and Max subscribers can access the red…

  19. So most people know now, If you properly configure your AGENTS and SKILLS files, you can get massive gains in how an AI agent performs. But the opposite is also true.

  20. event

    Model Context Protocol
    188 items

    Anthropic has launched free AI courses and disclosed a critical vulnerability in its Model Context Protocol (MCP) affecting 200,000 servers, while the protocol continues to be explored for its potential in improving coding agent performance. New features like self-hosted sandboxes for Claude Managed Agents are also being introduced.

  21. i know i have been posting a lot on this subreddit, but i am kinda fully interested into building my software rn and i am running into lots of issues, so please HELP. SO Running agents unattended and the real cost isn't bad output, it's th…

  22. Thurbox Run several coding agents at once — Claude Code, Codex, Antigravity, opencode, aider, or any CLI you describe yourself — side by side in one terminal. Each gets its own persistent tmux session and its own git worktree, so they neve…

  23. could not extract summary

  24. I tried computer-use agents with Astra and Fable 5.1 to automate tasks on my PC. The problem was simple: the cost didn’t make sense for the repetitive browser work I wanted to delegate.

  25. could not extract summary

  26. kibana-agent Read-only Kibana/ES CLI for AI agents. Queries Elasticsearch through Kibana's console proxy API.

  27. Change the way agents work with Postgres by treating WAL as a durable source of truth by Cassie Murray and Carlota Soto Agents that interact with a traditional OLTP database often create bottlenecks at the storage layer. New deployments, c…

  28. I've been running Claude code and Codex in parallel daily like Claude Code on the frontend and codex on the backend. The problem was keeping the agents in parallel like every switch meant reexplaining context, reuploading files and to mant…

  29. For reference, SFX and and 3D models, as well as animations were all fully authored by Claude. My part was visualizing them in pain or handrawing them on paper and sending him images to convey the idea.