@amitkumar0331 · X

@iveyzen · X

鹿友AI上线了 最贵的AI中转站，只因为想创造稳定fable-5访问专线

@WebWrightAI · X

fran-mora · HN

I gave 5 AI coding agents one prompt: grow a banana plant through its whole life in three.js: sprout, leaves, flower, fruit, rot, then pups that restart the loop. It's deceptively simple and yet very hard to get right from procedural code: you have to write working three.js and understand how the plant is actually built; how it hangs, ages and decays. Get the biology wrong and the code renders something weird. These are agents, not bare models (Claude Code and Codex for now). They can use tools, including playwright to check their work and improve it.

@amiuchat · X

我做了一个桌面工具：Token Switch。 给 Codex / Claude Code / OpenCode 重度用户统一管理 Provider、模型、Token、Base URL、连通测试和消耗统计。 多个 Agent，一个 Token 工作台。欢迎试用，也欢迎吐槽你最烦的配置问题。

@frank_brsrk · X

reasoning tools for ai agents

@computer_agents · X

@iampeersky · X

is blazingly fast LLM generated code slop detector & clean up cli tool that installs as git pre-commit hook to save your git history from sloppppp

niklio · HN

You write 8 text completions and open models score how predictable each word was too them. Predictable => clanker. You can share results with your friends. The scoring checks every word you write against the model's logprobs. Right now I'm using Llama3.1, Deepseek v3 and Qwen3 to keep costs low. I tried to calibrate it so other models (chatgpt/claude) score 100% and interesting human responses score in the 10-30% range. Totally free, no signup

@shokhkarim1212 · X

@_m3rl1n