AI圈报教程 / 实战普通
Home-made Evals Just Taste Better
内容摘要
Note This is a shortened version of the original writing on kig.re — my personal blog. Here is...
相关推荐
综合分类、实体标签与信息来源匹配
教程 / 实战普通
Will AI Make Abstractions Obsolete?
As AI agents become increasingly capable, I've noticed a growing number of discussions questioning...
信息来源：DEV Community
教程 / 实战普通
Before you open Claude Code in a cloned repo, check its .claude/skills folder
Before you open Claude Code in a repo you cloned, look inside its .claude/skills/ folder. Read the...
信息来源：DEV Community · Claude
教程 / 实战普通
Replaying 147 Claude Code & Codex Sessions to Tune autoCompactWindow (and Catching a 3x Token Overcount)
My first script for checking local Claude Code token usage summed raw assistant lines in...
信息来源：DEV Community · Claude
教程 / 实战普通
One word in a chat took my shop offline. I built it that way.
The word was go. A bot in the on-call room had just told me the deploy caused the outage. It quoted...
信息来源：DEV Community
教程 / 实战普通
CallBook: a prediction you can't take back, remembered in Walrus and proven through Tatum
Everyone says they called it. Almost nobody can show they said it first. That is the problem...
信息来源：DEV Community
教程 / 实战普通
Congrats to the Kaggle Benchmarking Challenge Winners!
A few weeks ago we teamed up with Kaggle to ask a deceptively simple question: how do you actually...
信息来源：DEV Community
教程 / 实战普通
Coding Agents Are Growing Memories. If You Can't git diff It, Don't Trust It
Codex 0.165 made memory v2 the default for fresh profiles; Claude Code's auto memory is on by default. Neither shows up in a PR. Agent rules belong in git.
信息来源：DEV Community · Claude
教程 / 实战普通
The Pace Doesn't Ship
You finish the engagement on a Friday. The repo is clean, the pipeline is green, the agent guides in...
信息来源：DEV Community
教程 / 实战普通
Prompt Pad: What Are Open-Weight Models?
When people call an AI model "open," they don't always mean the same thing. On this week's Prompt...
信息来源：DEV Community
教程 / 实战普通
Threat-model an agent that can spend or sign
Map what a compromised prompt, model, or host lets an attacker sign or steal, then choose a control tier.
信息来源：DEV Community