Hister ships a self-hosted full-text search index you fully control

Hister is a self-hosted, free-software search tool that builds a private full-content index out of the web pages, files and browser history a user chooses to keep, storing that index and the extracted document content entirely on a server the user configures rather than in a cloud service. Its source code is public and licensed under AGPLv3, and it runs either as a single local binary or as a shared server with user-scoped access backed by SQLite or PostgreSQL.

Content gets into the index in several ways: a browser extension that indexes pages as they are visited, a file watcher for local folders, an import of existing browser history, or a crawler pointed at a site. Hister then extracts the parts of each document that matter and builds a full-text index that supports field filters, quoted phrases, wildcards, negation, date ranges and user-defined query aliases, alongside skip and priority rules that shape what gets indexed and how it ranks. Search results come with a stored, readable preview of the original content shown alongside the results, so a user can look inside a page without leaving the search view.

The indexed data can be queried through a web interface, a terminal client, a command-line tool, an HTTP API, or an MCP server that lets an AI assistant retrieve results directly. The project states that the server itself carries no telemetry, makes no external requests, and does not phone home or report what a user searches for, meaning a complete personal setup can run entirely on one local machine with clients sending indexed content only to the configured Hister server. Two features are explicitly opt-in and reach outside the local setup: semantic search, which sends text to an embeddings endpoint the user configures, and browser extensions retrieving page favicons; both are under the user's control as to whether and where they run. Additional stated capabilities include support for multiple crawler backends, language-specific indexes, content versioning that can retain earlier versions of a document, ownership rules, and configurable content extractors for structured data from supported file formats and websites.

Key facts

Hister is free software licensed under AGPLv3, with source available on GitHub, and runs as a self-hosted single binary or a shared server with user-scoped access via SQLite or PostgreSQL.
It indexes content collected through a browser extension, local folder watching, browser history import, or site crawling, then extracts and full-text indexes the material on a server the user configures.
Search supports field filters, quoted phrases, wildcards, negation, date ranges and query aliases, and the index is reachable via web, terminal, CLI, HTTP API, or an MCP server for AI assistant access.
The project states the server carries no telemetry, makes no external requests, and does not report what a user searches for; a full personal setup can run on one local machine.
Optional semantic search sends text to a user-configured embeddings endpoint, and browser extensions may fetch page favicons; both connections are opt-in and user-controlled.

Why it matters

Search tools that index a person's browsing history and files typically run through a vendor's cloud, which means the content of what someone reads and keeps passes through a third party. Hister's pitch is to keep that entire pipeline, collection, indexing, storage and search, on infrastructure the user owns, with the source published under AGPLv3 so the claims about no telemetry and no external requests can be checked against the code rather than taken on trust.

Who it affects

The tool targets individuals and teams who want to retain and later search what they have already found useful: saved pages, local documents, and browser history, plus anyone building AI-assisted workflows who wants an assistant to retrieve from a personal knowledge base through MCP rather than a public search API. A shared-server mode with user-scoped access on SQLite or PostgreSQL extends this to small teams rather than only single users.

How to use it

Hister can be installed as a single local binary for a personal setup, or deployed as a shared server backed by SQLite or PostgreSQL for multiple users with scoped access. Content enters the index via a browser extension, a file watcher on local folders, a browser history import, or a crawler; once indexed, it is searchable from a web interface, a terminal client, a CLI, an HTTP API, or through an MCP server for AI assistant integration. The source text does not state pricing or a business model, and none should be inferred.

How solid is it

The claims in this piece come from Hister's own landing page and documentation, not from independent testing or benchmarks; the source text gives no performance figures, index size limits, or scale numbers, and makes no comparison to competing tools. The software being AGPLv3 and publicly sourced on GitHub means its no-telemetry and local-storage claims are at least auditable in principle, but the source material itself is a vendor description rather than a third-party review.

Risks and caveats

As with any self-hosted tool, the burden of running, securing and backing up the server shifts to the user, and the source text gives no indication of how mature or actively maintained the project is, nor its version history or release date. The two features that reach outside the local setup, optional semantic search and browser-extension favicon fetches, still involve external network calls even though they are opt-in, so a user who wants a fully closed setup needs to leave them off.

“Hister turns the pages you visit and the files you keep into a private, full content search index that you control.”

— Hister landing page