知识库

Eye(FR) → Mill(skill) → Library(/app)

← 返回列表
Hacker News · 2026-05-13 · 已成文 · 来源 file

What is your AI agent constantly doing wrong? — 来自收藏队列的工程/Agent 相关条目,沉淀为可检索 claim 卡。

为什么重要

收藏条目进入 AI 处理集后必须补写,否则 star 与知识文断档;本卡把收藏接回可检索结论。

正文

What is your AI agent constantly doing wrong? — 来自收藏队列的工程/Agent 相关条目,沉淀为可检索 claim 卡。

Claim

What is your AI agent constantly doing wrong? — 来自收藏队列的工程/Agent 相关条目,沉淀为可检索 claim 卡。

Why it matters

收藏条目进入 AI 处理集后必须补写,否则 star 与知识文断档;本卡把收藏接回可检索结论。

Summary

收藏(FR favorite)无 ka 条目的 Mode B 补写。1:1 挂在同 URL Vault item。 正文四段给出可执行动作;证据含原文与 HN/讨论入口(若可得)。

Actions

  • (none)

Evidence

  • (none)

Caveats

  • (none)

Research queries

  • (none)

Body

正文

背景

FR 收藏但无 ka-v0,导致处理集与知识库断档。

机制

Mode B:同 item 写入围栏 ka-v0 + 正文四段;FR mark_read 对齐;不 bulk 清未读。

取舍

覆盖收藏缺口 vs 单篇深度;本批优先接通链路,深度可后续杂志化。

动作

核原文 → 合并同主题 → 纳入工作台检索。

结合的源文章

主源
What is your AI agent constantly doing wrong?
打开原文 ↗

原文快照

展开 / 收起快照

My colleague and I have found AI is consistently falling short on the following things. I'd be interested to know whether this is a prompting issue or a model issue (we use claude 4.7):

- comments being too verbose, and making reference to the previous code. Also just using phrases that no human would ever use (e.g. saying 'gate', 'belt-and-braces' a lot)

- over-using default arguments in method definitions

- failing to see the bigger picture in terms of DRYness, code organisation

- way too much defensive coding like handling edge cases that would never happen, adding fallbacks for everything


Comments URL: https://news.ycombinator.com/item?id=48120148

Points: 1

# Comments: 0