## Summary
umerkay/mike
umerkay bolts a live web-research arm onto Mike, so the assistant can read the open web on demand.
This is a lean fork. umerkay took Mike and gave it a way to reach out to the live web instead of working only from what it already knows. The assistant can now pull a single page, ask a question and get an answer backed by real sources, or fan out across a whole site and crawl it.
The plumbing behind this leans on Olostep, a hosted service that handles the scraping and crawling. That means the fork isn't reinventing a web-research stack - it's wiring Mike into one that already exists and exposing a handful of endpoints on top. For a legal-tech reader, the interesting bit is the shape of it: an assistant that can go fetch and cite current material rather than answer from a frozen snapshot.
There's no rebrand here and no stated niche. It's one focused piece of work by a single author, and the direction - if it continues - points at grounding Mike's answers in live external content.
What's in it
- Single-page fetch Point the assistant at a URL and have it read the page.
- Sourced Q&A over the web Ask a question and get an answer backed by content pulled from the open web.
- Site mapping and crawling Walk a whole site rather than a single page when the answer isn't in one place.
- Hosted research backend The heavy lifting runs through Olostep, so the fork wires into an existing scraping-and-crawling service instead of building one.
Direction
searchintegration
Activity
## Summary
A proposed integration would let the assistant pull in live content from across the internet, not just the files a user uploads.
Four new endpoints let the assistant pull live content off the open web - scrape a page, ask a question and get a sourced answer, or map and crawl a whole site.
Threads of work (detailed view)
umerkay wants Mike to reach out to the open web
A proposed integration would let the assistant pull in live content from across the internet, not just the files a user uploads.
umerkay gives Mike a web-research arm
Four new endpoints let the assistant pull live content off the open web - scrape a page, ask a question and get a sourced answer, or map and crawl a whole site.
umerkay adds Olostep web research endpoints behind optional dependency
Four backend POST endpoints at `/api/research` - scrape a URL to Markdown, run an AI Q&A against a page, start a multi-page crawl, or discover site URLs - all gated behind an optional `olostep` SDK install and `OLOSTEP_API_KEY` env var. Absent the key, every call returns an error string; nothing breaks on boot.
Pull requests (detailed view)
⛔ Closed without merge (2)
by
umerkay · opened 3mo ago · closed 2mo ago by
umerkay · opened 3mo ago · closed 2mo ago