TE
科技回声
首页24小时热榜最新最佳问答展示工作
GitHubTwitter
首页

科技回声

基于 Next.js 构建的科技新闻平台,提供全球科技新闻和讨论内容。

GitHubTwitter

首页

首页最新最佳问答展示工作

资源链接

HackerNews API原版 HackerNewsNext.js

© 2025 科技回声. 版权所有。

Show HN: I made PromptWatch – LangChain tracing on steroids

10 点作者 ju-bezdek大约 2 年前
As I was building my first few chains, I noticed how difficult it was to understand what was going on, why GPT was responding the way it was, and how I could improve it. Tinkering with my prompts was so tedious - I had to either update the prompt template and rerun the whole chain, or grab the actual prompt, take it to OpenAI playground, tweak it there, and then make a template out of it again. That&#x27;s why I decided to build something better!<p>PromptWatch allows you to tune your prompt templates in the browser in a format appropriate for LangChain, decoupling template from parameters data.<p>It also allows you to track your sessions and costs per your project and tenant, track template versions and more.<p>Best of all, it&#x27;s completely free! I would be thrilled to hear any feedback you have. Thanks so much for checking it out!

1 comment

anentropic大约 2 年前
I&#x27;d like clarity on how it works<p>i.e. it&#x27;s &quot;completely free&quot; but also involves a signup and a login<p>what data is going via the PromptWatch servers?