No More Scrolling Through Group Chats, But Now There's a 20,000-Word Daily Digest Every Day.

A developer named Rion Wu built a personal WeChat intelligence repository: it reads chat logs via WeChat CLI and automatically summarizes key information from the past 24 hours. His original goal was to escape information overload from group chats. It worked—he no longer has to scroll through endless group chats, but instead he now has to read a 20,000-word daily digest every single day. He joked, "I really walked right into this myself."

Instead of piling more rules into the prompt, he wrapped his frequently used research workflow into a Project Research Skill, then had Codex search and compare existing public solutions via AnySearch. After running the process, he realized all requirements had been crammed into one single "one-size-fits-all" report, and no matter how long the digest was, it just wasn't useful.

So he decided to split the intelligence repository into five separate access points:

- Today's Overview: What happened, and what matters most

- Topic Search: Quickly locate a specific message, link or discussion

- Contacts: Fill in conversation background and relationship context

- Reply Suggestions: Draft context-aware responses

- Deal Radar: Specifically filter for collaboration and earning opportunities

Deal Radar is no longer shown by default, and only appears as an optional entry. You check the overview when you want a full recap, search directly when you're looking for specific materials, pull up context when you need to reach someone, and generate suggestions only when you need help replying.

This revised version is set to be open-sourced this week. The full version includes the following features:

1. Summarize 24-hour chat history via WeChat CLI and provide an overview of action items

2. Auto-summarize important group chat messages from the past 24 hours

3. New interactive HTML daily digest, more intuitive than Markdown

4. New daily digest for key contacts, with built-in reply suggestions

5. Cross-group link indexing to automatically identify collaboration opportunities

6. Customizable time range: 1 day, 1 week, 1 month, can summarize any existing chat history





The key tool that enabled this optimization is AnySearch, a search infrastructure built for AI Agents. Rion shared his workflow: when starting a new project or optimizing an existing one, instead of having the agent build everything from scratch, he first uses AnySearch to research existing public information, draws on proven solutions, then adapts them to his own needs. This avoids a lot of wasted work and lets you build an MVP quickly.

AnySearch natively supports Skill, MCP and API integration, and also allows anonymous usage with no tracking and zero telemetry. Privacy boundaries are clearly defined: WeChat chats are always processed locally via CLI, and AnySearch only handles research on public information and never accesses private chat content.

The free tier offers 1000 requests per day, and verified students and developers get 2000 requests per day. You can register and claim a free API Key on the official website.

This update has sparked some discussion. One netizen commented, "Prompts only solve how to do things, but agents are solving who does the work," and many agree that splitting into five entry points is far more practical than a single overstuffed report. Another user wrote, "This is way more useful than the paid course I bought."

The repository is already up on GitHub, and the official release will be open-sourced this week. Many people have been waiting for it, with one user joking, "I've clicked your avatar so many times checking for updates." It looks like this won't be another delayed release.

AnySearch Official Website: https://anysearch.com/home?utm_source=kol_Rionwu&utm_medium=referral&utm_campaign=kol_08a

GitHub: https://github.com/anysearch-ai