Hister Review: A Self-Hosted Search Engine for Everything You Have Already Read
Bookmarks rot, browser history search only matches titles and URLs, and the page you actually need is always the one you forgot to save. Hister attacks that specific failure: it indexes the full content of the pages you visit and the files you keep into a private search index that runs on a server you control. The project reached the Hacker News front page in August 2026 with 423 points and 94 comments, which is a strong signal that "my own search engine for things I already read" is a widely felt itch. This review is based on Hister's public documentation and site as of August 2026. Hister is free software under the AGPLv3 license, so there is nothing to buy โ the real cost is running and maintaining the server. The question this review answers is whether that trade is worth it, and for whom.Quick Verdict
| Aspect | Assessment |
|---|---|
| What it is | Self-hosted, full-content search index of pages you visit and files you keep |
| Price | Free software (AGPLv3); you supply the machine it runs on |
| Best for | Developers, researchers, and privacy-conscious users who want searchable recall without a cloud service |
| Standout feature | Full-content indexing with stored previews, queryable from web, terminal, CLI, HTTP API, and MCP |
| Main caveat | You operate the server; indexing everything you read creates its own sensitive archive to protect |
What Hister Actually Does
Hister's own summary is precise: it "turns the pages you visit and the files you keep into a private, full content search index that you control." Three design decisions follow from that sentence. First, full content, not metadata. Bookmarks and history search fail because they index titles and URLs. Hister extracts and indexes the text inside the page or file, then stores a readable preview alongside the index. When you search, you can open the stored preview next to the results and get back to the exact idea you remembered โ even if the original page later changes or disappears. Second, your server, your rules. The index, stored page content, and rules stay on the Hister server you configure. The project states the server has no telemetry and makes no external requests, and a complete personal setup can run on one local machine. The source is public and licensed AGPLv3, so the privacy model is auditable rather than a promise. Third, many doors into one index. The same index is reachable through a web UI, the terminal, a command-line client, an HTTP API, and MCP โ which means an AI assistant can retrieve from your personal index as a tool. That last door is what makes Hister feel like a 2026 product rather than a 2016 one.How Collection Works
An index is only as good as what goes into it. Hister offers four intake paths:- Browser extension: indexes pages as you visit them. A Chrome Web Store listing is linked from hister.org, and the extension sends content only to the Hister server you configure.
- Local file watching: point it at folders (notes, documents) and it indexes the files you keep.
- History import: backfill from your existing browser history instead of starting from zero.
- Site crawler: deliberately index an entire site you want searchable offline.
Search Quality: The Query Language
Full-text search tools live or die on whether you can narrow precisely when the index grows. Hister supports field filters, quoted phrases, wildcards, negation, date ranges, and user-defined query aliases. That is the correct checklist for a personal index that will eventually hold tens of thousands of documents: recall queries ("that privacy article from last month") need date ranges and phrases; precision queries ("only in my notes, not crawled pages") need field filters. Semantic search exists but is explicitly optional โ and the documentation is honest about the trade: enabling it sends text to whatever embeddings endpoint you configure. You choose whether and where that connection runs. For a privacy-positioned tool, making the cloud-shaped feature opt-in and endpoint-agnostic is the right default.The Privacy Model, Read Closely
The headline claims โ no telemetry, no mandatory cloud, auditable AGPLv3 source โ are the baseline you should demand from any tool in this category, and Hister meets them on paper. Two nuances deserve attention before you commit: 1. You are building a sensitive archive. A full-content index of everything you read is more revealing than a browser history. Self-hosting removes the vendor from the threat model, but it makes you the security team: disk encryption, access control on the web UI, and sensible indexing exclusions are your job.2. Optional features can phone out. Semantic search uses the embeddings endpoint you configure; the browser extension may retrieve favicons. Neither is hidden, and both are your choice โ but "self-hosted" describes the default posture, not a guarantee about every optional path.
Hister vs the Alternatives You Already Have
- Browser history search: matches titles and URLs only, loses content when pages change, and is tied to one browser. Hister indexes content, keeps stored previews, and unifies sources.
- Bookmarks / read-it-later apps: capture what you remembered to save. Hister's extension captures as you browse, which is the difference between a curated shelf and a searchable memory.
- Commercial "recall everything" tools: typically cloud-dependent, subscription-priced, and opaque about where your data lives. Hister trades their convenience for your operational effort and full control.