Big Data Digest, authorized reprint from Digital Life Kazx
Author: Kazx
Yesterday, I made my AIHOT, an AI hotspot monitoring website, freely available to everyone.
It has been well-received by many friends.
I never expected that on the first day, it would surpass 100,000 unique visitors, with page views exceeding 600,000. There were hardly any negative reviews or bugs, and the feedback has been very positive, so I finally breathed a long sigh of relief.
Yesterday, the top two requests I received were: first, the dark mode was difficult to read, and whether a light mode could be added. That was indeed my oversight, so I spent over an hour yesterday morning urgently implementing it, and it was live by noon yesterday.
The second request was whether skills, APIs, or RSS feeds could be added.
I also received a nudge from Mr. Ruan, so it had to be done 
.
This is the AI era, after all. Having only a website felt too rigid. So when I got home from work last night, I decided to fire up my AI and start coding to fulfill everyone's requests.
And after an all-nighter, I've finally finished developing everything.
Today, I think I can open up my AIHOT to all agent users as well.
And as usual, it's completely free for everyone.
Here's the URL: https://aihot.virxact.com
Go to the AIHOT main site and click "Agent Access" on the left, and you'll see it.
Three access methods are currently available:
Skill, RSS, and API, each corresponding to different needs.
I've also opened up the data that I thought could be shared to the greatest extent possible.
Let's focus on the AIHOT Skill, which is probably the most requested and the most crucial thing in the AI era.
I won't go into lengthy explanations about Skills. In short, they are capability packages for agents. If you're unfamiliar, you can search for "Skills" on my public account; I've written a lot about them.
You can install it using any agent that supports the Skill protocol, such as Claude Code, Codex, OpenCode, OpenClaw, Hermers, and others.
The function of this AIHOT.skill is quite simple: it allows your agent to directly read some of the data from my AIHOT website, embedding it into your workflow.
Installation is also very simple, with just one line, and I've hosted it on my server, so no proxy is needed:
Help me install this skill: https://aihot.virxact.com/aihot-skill/Of course, as usual, I've also uploaded this skill to my Skills collection on GitHub. If you want to use the GitHub source, you can find it here: https://github.com/KKKKhazix/khazix-skills
After installing the AIHOT Skill, you won't need to open a browser, refresh a website, or even think about what happened in the AI world today.
You just need to say one sentence to your agent.
Let me elaborate a bit on what it can do.
First, the AI Daily Brief.
AIHOT automatically generates a daily AI brief every day at 8:00 AM Beijing time. This brief is the result of my system crawling, filtering, deduplicating, scoring, and categorizing data from hundreds of sources.
For example, when you wake up and open Claude Code or OpenClaw, simply say, "Give me today's AI daily brief," and it will automatically trigger the AIHOT Skill, pull down the day's brief, and organize it into a concise rundown, right in front of you.
The daily brief is divided into five sections: Model Releases/Updates, Product Releases/Updates, Industry News, Research Papers, and Tips & Opinions. Each item has a Chinese title, a one-line summary, the source, and a link to the original article, which you can click to read if you're interested.
In 30 seconds, you'll know what happened in the entire AI industry yesterday.
And you're not just limited to today's brief; you can also check previous days'.
For instance, you can tell your agent, "Show me the AI daily brief from May 6th," and it will directly pull up that day's brief. I personally love this scenario and find it perfect for weekends.
I believe many people don't check AI news much on weekends and catch up on Sunday evening or Monday morning. In that case, you can simply say on Monday morning, "Summarize the AI daily briefs from the last three days for me," and it will pull them all up quickly.
The second capability is the Curated Feed mode.
The difference between the Curated Feed and the daily brief is that the brief is a packaged product organized by sections, like a newspaper, with a maximum entry limit. The Curated Feed, on the other hand, consists of noteworthy pieces of information selected by AI from all sources, presented in a chronological feed format.
The Curated Feed is the core mode of AIHOT. Unless you explicitly request the daily brief or all entries, the agent will default to answering based on the curated data source.
This really depends on your needs. If you just want a quick morning scan of major events, use the daily brief. If you want to ensure you don't miss any high-quality entries, go for the Curated Feed.
The third capability is querying by time window or category.
Sometimes you might feel the information from the daily brief or curated feed isn't enough and you want to see all activities in a specific direction.
Above that level, there is actually the full volume of all AI-related information, which might not have been selected for the curated feed to protect your attention, but that doesn't mean it's not valuable.
At this point, you can tell your agent something like, "Show me all messages and list all new model releases," and it will pull up all relevant entries for AI models. Or, "Show me all AI product releases from the last 3 days," and it will filter everything in that product direction.
The categories are the same as the daily brief, supporting these five: Model Releases, Product Releases, Industry News, Research Papers, and Tips & Opinions.
Of course, you can also specify a time range, like "What big news has happened in the AI industry in the past 24 hours?"
In about a minute, you can get very detailed and accurate information.
Naturally, by default, it uses curated information to answer, protecting our collective attention spans.
The maximum supported time window is 7 days, as the data volume before that becomes too large. It's also to protect my humble potato server—I'm seriously afraid a huge volume would crash it…
The fourth capability is keyword searching.
I've also integrated a search function to support basic queries, such as recent updates for a specific product or new models released by OpenAI.
This is a great opportunity to show a comparison of information timeliness with and without the skill.
For example, if I ask, "What has OpenAI released recently?"
With the skill installed, the answer is very detailed and comprehensive, even capturing a voice model released just after midnight.
And each entry has a source link you can jump to if needed.
Without the skill installed, this is what it looks like:
It can be a bit off and missing key information.
That's about it.
As for the output format, I only did basic markdown, because I felt it was best to just provide a data source. Since the skills can be freely modified, you can format the output however you like or embed it into your workflows—I'll leave that optimization to you.
I hope the AIHOT skill gives your agent an extra pair of eyes to track the news across the entire AI industry for you.
Next, let's talk about RSS.
This is really for those friends who don't use agents but use RSS readers daily. Many tech industry veterans are accustomed to using tools like Feedly, Inoreader, and NetNewsWire to manage their information feeds. For them, RSS is the most natural way to connect.
I've made three feeds available: Curated Updates, All AI Updates, and the AI Daily Brief. Just pick the ones you need and subscribe. The feed addresses are all visible on the Agent Access page, ready for one-click copying.
And then there's the API. Many friends mentioned wanting to integrate the data into their company's internal systems or other tools. After thinking it over, I decided to make it available to everyone.
If you're building an AI-related product and want to integrate AIHOT data into your own system, you can directly call the API.
I've made the complete OpenAPI specification documentation as detailed as possible, with the help of my AI. But I have to flag a risk here: I've never personally opened an API before, and I don't really understand the technicalities well. This part was entirely handled by my agent after I described my requirements and some risk-control issues, so I'm genuinely not confident about the API.
If you encounter bugs during the integration process, and after checking the docs you find it's a non-standard error, please submit a feedback report, and I'll have my agent fix it. 😅
Three access methods, covering three different groups of people.
In short, my only hope is that this tool is useful to you. Regardless of how you prefer to get your information, I hope AIHOT can provide a little bit of help.
I also hope that I can leave a small, personal mark on this internet.
People who hit 'Like' have all become better looking!