0xKong

[0xKong @gm5gn]

Claude Code 结合 DeepSeek V4 Pro 使用，在 settings.json 中配置了 "errortLevel": "max"，但是不生效，每次都要在窗口中切换，很麻烦。后来查阅了文档，发现 errortLevel 的属性值 max 是不支持的，但是可以用 "CLUADE_CODE_EFFORT_LEVEL": "max"。又学习了一个小技巧。。。

[0xKong @gm5gn]

@object_nullll 能否分享好的 eval 案例？

[0xKong @gm5gn]

好马易得，善骑者胜。

[Tomasz Tunguz @ttunguz]

Model capabilities are converging.

The best riders win.

https://t.co/3T2B3aqVs1

[0xKong @gm5gn]

@dotey 之前用 OpenSpec 结合 Claude Opus 写proposal，然后 Superpowers 结合 DeepSeek V4 Pro 执行和测试，发现写代码效果还可以。由于 AI 写代码太快了，我的审核、验证太慢，导致逐渐偏离需求。

其实 AI 已经能写出很好的代码了，但是审核、验证还缺乏完善的辅助工具，让我们来匹配 AI 生成的速度。

[0xKong @gm5gn]

@9hills 之前调研过，基本上推荐这三个：
https://t.co/ps0M6tVRS0
https://t.co/vEPPCbjJqa
https://t.co/6pWlSLjiJE

记得安装它们对应的 skills，这样 AI 才能写出符合要求的代码。
https://t.co/lHqP6r3KgZ
https://t.co/zLsZ6OCfCD
https://t.co/CiaDkblQq6

[0xKong @gm5gn]

Antigravity 2.0 配置自定义 MCP （非官方 MCP），以 LangChain Docs MCP 为例。
~/.gemini/config/mcp_config.json
```json
{
"mcpServers": {
"docs-langchain": {
"serverUrl": "https://t.co/IXFBxH69uV"
}
}
}
```
#Antigravity #MCP

[0xKong @gm5gn]

@imwsl90 差点就以为是真的了。。。

大家还是自己多核实下，下面是官网。
https://t.co/YXOK3z1Vj2

[0xKong @gm5gn]

@manateelazycat 快半年了，还不更新。再也不年费订阅了🥹

[0xKong @gm5gn]

@lyricwai 一线的和坐办公室的还是有很大区别的，不亲身体验的话，很难共情

[0xKong @gm5gn]

@arkuy99 好办法

[0xKong @gm5gn]

@skywind3000 目前是 OpenSpec 做规划（propsal、spec、task）、验收、归档，Superpowers 根据 OpenSpec 的文档来编写更详细的计划并执行任务（TDD、Spec Review、Code Review），执行完成任务后同步 OpenSpec 的 task。好处是都可以依据文档上下文来新开会话执行，坏处就是太费 token 了。

[0xKong @gm5gn]

@ZeroZ_JQ 也可用别名方式，更简洁易记

alias claude-ds='claude --settings ~/.claude/settings-deepseek.json'

[0xKong @gm5gn]

@ZeroZ_JQ claude --settings ~/.claude/settings-deepseek.json

--settings 指定配置文件模式，可以在不同窗口使用不同供应商模型

Claude Code 配置 DeepSeek：
https://t.co/VUjxRxLe5F

[0xKong @gm5gn]

@dotey 哇 可以可以 之前 nano banana pro 试了好久，效果都不理想，这张效果很棒。

[0xKong @gm5gn]

@Ajayfizzy both

[0xKong @gm5gn]

@colinwu 第三条尤为重要，ai新产品一发布，就刷新认知。再加上网络上充斥着夸张、贩卖焦虑的信息。小小的心灵，每时每刻都在震撼，非常需要强大的身体和心理来消化这一切。

[0xKong @gm5gn]

@dotey 感谢宝玉老师系统全面的分析。太及时了，最近在作个人软件，token有时消耗慢有时消耗快，终于找到原因了。我是连续相关的任务在一个会话，但有时第二天继续在这个会话开发，一上来token就消耗了很多，看来得新开个会话才划算一些。

[0xKong @gm5gn]

@0xReggieJ @binghe 使用 OpenSpec 记录需求和任务，能记住一些关键信息。这个是存储到项目中的，跨平台的 cc codex antigravity 都可以用。

[0xKong @gm5gn]

@dotey AI时代，云厂商的另一个收入来源，面向结果交付。短时可能有炒作预期，但长期看，特别是随着模型能力的提升，这个方向还是很有前景。对于中小企业来说，买业务结果比买基础设施更划算。

[Andrej Karpathy @karpathy]

LLM Knowledge Bases

Something I'm finding very useful recently: using LLMs to build personal knowledge bases for various topics of research interest. In this way, a large fraction of my recent token throughput is going less into manipulating code, and more into manipulating knowledge (stored as markdown and images). The latest LLMs are quite good at it. So:

Data ingest:
I index source documents (articles, papers, repos, datasets, images, etc.) into a raw/ directory, then I use an LLM to incrementally "compile" a wiki, which is just a collection of .md files in a directory structure. The wiki includes summaries of all the data in raw/, backlinks, and then it categorizes data into concepts, writes articles for them, and links them all. To convert web articles into .md files I like to use the Obsidian Web Clipper extension, and then I also use a hotkey to download all the related images to local so that my LLM can easily reference them.

IDE:
I use Obsidian as the IDE "frontend" where I can view the raw data, the the compiled wiki, and the derived visualizations. Important to note that the LLM writes and maintains all of the data of the wiki, I rarely touch it directly. I've played with a few Obsidian plugins to render and view data in other ways (e.g. Marp for slides).

Q&A:
Where things get interesting is that once your wiki is big enough (e.g. mine on some recent research is ~100 articles and ~400K words), you can ask your LLM agent all kinds of complex questions against the wiki, and it will go off, research the answers, etc. I thought I had to reach for fancy RAG, but the LLM has been pretty good about auto-maintaining index files and brief summaries of all the documents and it reads all the important related data fairly easily at this ~small scale.

Output:
Instead of getting answers in text/terminal, I like to have it render markdown files for me, or slide shows (Marp format), or matplotlib images, all of which I then view again in Obsidian. You can imagine many other visual output formats depending on the query. Often, I end up "filing" the outputs back into the wiki to enhance it for further queries. So my own explorations and queries always "add up" in the knowledge base.

Linting:
I've run some LLM "health checks" over the wiki to e.g. find inconsistent data, impute missing data (with web searchers), find interesting connections for new article candidates, etc., to incrementally clean up the wiki and enhance its overall data integrity. The LLMs are quite good at suggesting further questions to ask and look into.

Extra tools:
I find myself developing additional tools to process the data, e.g. I vibe coded a small and naive search engine over the wiki, which I both use directly (in a web ui), but more often I want to hand it off to an LLM via CLI as a tool for larger queries.

Further explorations:
As the repo grows, the natural desire is to also think about synthetic data generation + finetuning to have your LLM "know" the data in its weights instead of just context windows.

TLDR: raw data from a given number of sources is collected, then compiled by an LLM into a .md wiki, then operated on by various CLIs by the LLM to do Q&A and to incrementally enhance the wiki, and all of it viewable in Obsidian. You rarely ever write or edit the wiki manually, it's the domain of the LLM. I think there is room here for an incredible new product instead of a hacky collection of scripts.