Why it exists

Everyone knows the feeling: "I read a really useful page about this a few weeks ago…" — followed by twenty minutes of re-Googling and digging through browser history. Bookmarks fail because they only store titles and URLs, not what the page actually said. Hister's thesis is that the antidote is a private memory: an index of the full content of the things you already found useful, stored on a server you control, searchable forever. Its demo says it best — 1,924 pages indexed, "privacy notes I read last month", found in 0.04 seconds.

Main features

• Full-content indexing — looks beyond bookmarks and filenames: it indexes the actual words inside every page and file you choose, and stores an extracted, readable preview alongside each result.
• Automatic collection — browser extensions (Chrome/Firefox) index pages as you visit them; file watchers index local folders; you can import browser history or crawl entire websites.
• Precision query language — field filters, quoted phrases, wildcards, negation, date ranges, priorities, and your own query aliases.
• Many ways back in — the same index is exposed through a web interface, terminal client, CLI, HTTP API and an MCP server (so Claude or any MCP-compatible AI assistant can retrieve your knowledge).
• Skip & priority rules — control what gets indexed and how it ranks; version tracking can retain earlier document content; sensitive-content checks are supported.
• Optional semantic search — sends text only to the embeddings endpoint you configure; everything else stays on your server.
• Lightweight deployment — a single binary can run locally; shared servers support user-scoped access with SQLite or PostgreSQL; Docker and Nix packages available.

Privacy model

• No telemetry — the server does not phone home or report what you search.
• No mandatory cloud — a complete personal setup runs on one local machine.
• Your chosen server — clients send indexed content only to the Hister server you configure.
• Auditable — public source under AGPLv3.

Who it's for

• Researchers, students and lifelong readers who remember reading something but not where.
• Privacy-conscious people who want a "second brain" without handing their reading history to a SaaS vendor.
• Developers and terminal users who want hister search instead of a web UI.
• Tinkerers wiring personal knowledge into local AI assistants via MCP.

How it compares

Useful links

• Website: https://hister.org/
• Live demo on hister.org
• Docs: https://hister.org/docs
• Extensions: Chrome & Firefox
• GitHub: https://github.com/asciimoo/hister · Codeberg mirror: https://codeberg.org/asciimoo/hister
• Community: Mastodon, Discord, IRCNet #hister

Notes

• One of the few personal-search projects designed MCP-first — a natural fit for the "AI assistant with local memory" workflow.
• The live counter on the homepage (pages indexed, search time in seconds) is a nice honesty signal.
• Supports language-aware indexes and multiple crawler backends for deeper site archiving.