Hister indexes your browsing history and files for private search

Hister is an open-source search engine you run yourself, built to index the full contents of the web pages you visit and the local files you choose to keep, so you can find that information again later instead of relying on a third-party search provider. It ships as a single server binary for Linux, macOS, and Windows: download it, make it executable, and start it with a listen command, or install it through Homebrew, Docker, or Nix. The server must keep running to index pages and to answer searches. Once it is up, the project's browser extensions for Firefox and Chrome save newly visited pages automatically, and the same content can be searched from a local web interface at http://127.0.0.1:4433, from a terminal client, or from an AI assistant connected through MCP (Model Context Protocol). Beyond automatic browsing capture, Hister can import existing browser history, crawl websites, index local directories, and import files directly. Search supports field filters, phrases, wildcards, negation, aliases, and result priorities, and an optional semantic search mode can find documents by meaning if the user configures a separate embeddings endpoint. The server also supports multiple users, keeping each person's documents and search results separate on a shared install. By default Hister collects no telemetry and does not sync to any cloud service: the browser extension sends indexed page content only to the Hister server the user has configured, with page favicons the one exception fetched from elsewhere. Turning on the optional semantic search sends document text to whichever embeddings endpoint the user picks, which is a separate, opt-in data flow the project's own documentation tells users to review before enabling. Building Hister from source requires Go 1.26, npm, and a C compiler for CGO dependencies, and the code is released under the AGPLv3 license or any later version. The submission reached the Hacker News front page and drew more than 500 points and over 140 comments within about half a day of posting.

Key facts

Hister is a self-hosted search engine that fully indexes the contents of visited web pages and local files, not just titles or URLs
It is searchable from a local web interface, a terminal client, or an AI assistant connected through MCP
By default it sends no telemetry and uses no cloud service; the browser extension only talks to the Hister server the user configures
An optional semantic search mode sends document text to a user-chosen embeddings endpoint, a separate opt-in data flow
The project is AGPLv3-licensed, installable via a downloaded binary, Homebrew, Docker, or Nix, and drew over 500 Hacker News points in about half a day

Why it matters

General search engines only see what is public. Hister targets the other half of a person's information: the pages they have already read and the files they already hold, kept in a private, full-text index instead of scattered across browser history and folders. Doing that locally, with no telemetry and no mandatory cloud service, is the part that distinguishes it from browser history search or cloud note tools that route the same content through a vendor's servers.

Who it affects

Anyone who wants to find something they already saw but cannot recall where; developers and researchers who accumulate documentation, papers, or web pages faster than bookmarks can track; and, through the MCP client, users who want an AI assistant to search their own past browsing and files rather than the open web. Multi-user support extends this to a small team sharing one server while keeping each person's index separate.

How to use it

Download the platform binary from the latest release, make it executable, and run it with a listen command to start the local server (default address http://127.0.0.1:4433); it can also be installed via Homebrew, Docker, or Nix. The server has to stay running to index pages and to search them. Installing the Firefox or Chrome extension then captures newly visited pages automatically; existing browser history, whole websites, local directories, and individual files can also be imported or crawled in. No configuration is required for a basic local setup. Building from source needs Go 1.26, npm, and a C compiler. The project is free and AGPLv3-licensed; no pricing or paid tier is mentioned.

How solid is it

The description comes from the project's own GitHub README and documentation, which lay out a quickstart, installation options across three platforms and three package managers, a full feature list, and separate pages on privacy and semantic-search configuration. No individual author, company, or team is credited in that text itself. The README gives no version number or release history for Hister, and no performance benchmarks, comparisons to other search tools, or user or download counts are included, so its maturity and how it holds up at scale are not established by the source.

Risks and caveats

Everything depends on a server the user keeps running: if it is down, nothing gets indexed and nothing can be searched. The privacy guarantee is a default, not an absolute: enabling the optional semantic search means sending document text to whatever embeddings endpoint is configured, which can be a third-party service, and the documentation itself says to review that configuration before turning it on. With no independent benchmarks, comparisons, or usage figures available, claims about index quality or scalability rest on the project's own description rather than outside verification.

“By default, Hister has no telemetry and no cloud sync.”

— Hister's project documentation