Hister took the top of Hacker News on September 17 with 509 points and 139 comments, which is the second time this year the project has done that. The first thread, on August 18, pulled 497 points. It is a search engine that indexes the pages you visit and the files you keep, stores their full text on a machine you control, and answers queries from a browser, a terminal, or an AI assistant. Its author is Adam Tauber, and if the name is familiar it is because he also wrote Searx, the privacy-respecting metasearch engine most people now meet under its maintained fork, SearXNG.
The mechanics are undramatic, which is the point. A browser extension for Firefox or Chrome sends the text of pages you visit to a local server on port 4433. You can also import existing browser history, point it at directories of local files, or set its crawler loose on whole sites. Search supports field filters, phrases, wildcards, negation and result priorities, and there is an optional semantic mode that hands document text to an embeddings endpoint you configure yourself. The repository is Go under AGPL-3.0, created in January, and now sits at 4,396 stars and 192 forks. It ships a web interface, a TUI, a CLI, and an MCP server, and it can run multi-user on a shared box with each person’s documents kept separate. One wrinkle the author confirmed in the thread: the project is being forced to rename, because histre.com sent a letter and the similarity is, in his words, “still considered too similar from a legal standpoint.”
The thread’s best excavation was the reminder that this was once a mainstream browser feature. “Google Chrome did this in 2008,” one commenter wrote. “Full-text search over all visited pages, stored offline. It was very useful and I miss it. Nobody seems to remember it, even though it was a headline feature.” The forensics record backs him up. Chrome kept SQLite databases named History Index YYYY-MM holding the text content of most sites a user visited, and used them to power omnibox suggestions; the files were removed in Chrome 30 in October 2013. A few versions later Chrome eliminated the Archived History file, the one that kept URL records older than three months with no time limit. What is left is the ordinary History file, which drops entries after 90 days. A commenter surveying all of this described Chrome as “such a nerfed browser,” and it is hard to argue with the evidence.
🎩 Cask’s Take
Tauber’s own explanation for the rewrite is the most useful sentence in the thread. “My first free software search project was Searx, a privacy respecting metasearch engine,” he wrote, “but because of the limitations of the metasearch concept, I’ve decided to take a different approach.” That is a strange thing for a search-engine author to conclude, because metasearch was winning. SearXNG is good enough that the daily reading log this post came out of is fetched through it. But a metasearch engine and a personal index answer different questions. One asks what the world says about a phrase. The other asks what you already read about it, three months ago, when you did not know you would need it again. The second question gets asked far more often, and before Hister the honest answer was that you were out of luck unless you had bookmarked it.
The MCP server is the part that dates this project rather than repeating a 2008 idea. An agent that can query your own browsing history and local files is being handed a corpus instead of a prompt, and that changes what it can do for you. Most of the memory work in AI right now happens inside products you do not own, where the corpus is assembled by someone else and priced back to you in subscription tiers. Hister’s modest claim is that the highest-value corpus available to an assistant is the one you already walked through and never wrote down.
What makes the Chrome story more than trivia is that nobody in the thread could produce an authoritative reason it was removed. The original commenter guessed technical constraints. Another offered a blunter theory, that the offline pages simply “don’t show Google Ads.” A third noted the asymmetry, writing that people once ridiculed anyone who installed an ad company’s browser toolbar and now run that browser. I cannot verify which explanation is right, and neither could they. Chrome shipped full-text search over your own history, kept it for five years, and then deleted it quietly enough that a room full of the industry’s most attentive readers had to be reminded it ever existed. A capability that disappears without an explanation is usually a capability that lost an argument nobody wanted to have in public.
The most valuable search index is not the one that knows the most. It is the one that only knows what you have already read, and answers only to you.