AI Services Market › Use cases › Research & Knowledge
Research & KnowledgeDone with custom-built or unnamed agent Oct 28, 2026
Research Xiaohongshu notes and distill them into skills
An agent drives Chrome to search keywords, open top-liked notes, read screenshots and distill findings into reusable skills in a knowledge base. · Oct 28, 2026
- What They Did
- The open-source rednote-bootstrap skill pack uses agent-browser, a CLI that controls Chrome. It splits a question into sub-dimensions and generates keyword searches. It reads the search-result snapshot, opens the top-liked notes, and screenshots each page. A vision model reads the text in the images. It then dedupes and cross-checks the findings into a sub-skill saved in a registry. The user scans a QR code to log in once.
- How It Went
- For the publishing-process question, it read 6 top notes (best had 4479 likes) and produced a full guide. A second skill, on account warm-up, came from 4 notes. The author calls the project early-stage.
- Worth Knowing
- A full research run takes roughly 5-10 minutes. Limits are 5 search rounds, 30 notes and 15 minutes, and the login state is saved to a local file for reuse.
Try it yourself with custom-built or unnamed agent
Open [platform, e.g. Xiaohongshu] and search for [keyword]. Open the top [number] most-liked notes, read each one including images, and note what they have in common: hooks, structure, format, and tone. Finish with a short reusable guide titled '[skill name]' that I can paste into my notes, and do not post, like, or comment on anything.