hister

Product Introduction

1. Definition: Hister is a self-hosted, open-source search engine and knowledge management platform. Technically, it is a full-text search index server that extracts, stores, and retrieves content from web pages and local files under user control.
2. Core Value Proposition: Hister exists to solve the problem of information loss and reliance on external services by creating a private, searchable memory of a user's digital content. Its primary value is enabling personal knowledge preservation and private full-content search without telemetry or cloud dependency.

Main Features

1. Multi-Source Content Ingestion: Hister aggregates content from diverse sources into a unified index. It works via a browser extension for real-time page indexing, a file watcher for local directories, a history import tool, and a configurable website crawler. This creates a comprehensive personal search index from visited pages, documents, and browsed history.
2. Advanced Full-Text Search Engine: The core is a powerful search interface supporting complex queries. It works using field filters (e.g., site:, type:), quoted phrases for exact matching, wildcards, negation operators, date ranges, and user-defined query aliases. This enables precise document retrieval beyond simple keyword matching.
3. Stored Content Previews & Contextual Reading: Hister doesn't just index metadata; it extracts and stores the primary content of documents. It works by saving a clean, readable preview of the indexed material, which is displayed alongside search results. This allows users to review information without revisiting the original source, ensuring access even if the original page goes offline.
4. Privacy-First Architecture & Deployment Flexibility: The system is designed with data sovereignty as a core principle. It works by running as a single binary or in a container (Docker/Nix), storing all data (index, extracted content, rules) on a user-controlled server. It supports both single-user (SQLite) and multi-user (PostgreSQL) deployments with zero telemetry and no mandatory external network calls.
5. Extensible Integration via MCP & API: Hister provides multiple access points to the same index. It works by exposing a standard HTTP API, a command-line interface (CLI), a web UI, a terminal client, and a Model Context Protocol (MCP) server. The MCP integration is key, allowing AI assistants and LLM-powered tools to securely query the private index as a data source.

Problems Solved

1. Pain Point: The "I know I saw it somewhere" problem—the inability to reliably refind useful web pages, local documents, or ideas encountered previously due to reliance on browser history, bookmarks, or memory.
2. Target Audience: Privacy-conscious individuals, researchers, academics, journalists, software developers, knowledge workers, and anyone who accumulates a large volume of digital reading material and files and needs to organize and retrieve it reliably.
3. Use Cases: A researcher building a literature review can automatically index all visited academic papers and search their full text later. A developer can index internal documentation and error-solving forum posts for offline reference. An investigator can crawl and preserve a snapshot of a website for secure, searchable analysis.

Unique Advantages

1. Differentiation: Unlike cloud-based bookmark managers (Pocket, Raindrop.io) or desktop search tools (Everything, Spotlight), Hister combines full-content indexing, source-agnostic ingestion, and a self-hosted, private server model. Unlike public search engines, it searches only content the user has explicitly encountered or owns.
2. Key Innovation: The integration of a robust, personal full-text search engine with a privacy-by-design architecture and the Model Context Protocol (MCP). This transforms a private index into a retrievable memory for both humans and AI assistants, bridging personal knowledge management and modern AI workflows without data leakage.

Frequently Asked Questions (FAQ)

1. Is Hister a replacement for Google Search? No, Hister is a complementary personal search engine for content you have already visited or own. It searches your private index, not the live web, making it ideal for refinding information, not discovering new public information.
2. How does Hister handle my privacy and data? Hister is a self-hosted search engine with a strict no-telemetry policy. All data—indexed content, search queries, and rules—resides solely on the server you deploy. The software is open-source (AGPLv3) for full auditability.
3. Can I use Hister to search the content of PDFs and documents? Yes, Hister includes content extractors for common file formats. It can index the textual content of PDFs, Markdown files, and other documents from watched folders, making their full text searchable.
4. What is required to run Hister? You need a machine (local computer, VPS, or home server) to run the Hister server binary. The setup can be minimal (a single binary with SQLite) or scaled (Docker with PostgreSQL). Browser extensions and clients then connect to your server's address.
5. What is MCP support and why is it important? MCP (Model Context Protocol) allows AI coding assistants (like Claude Code or Cursor) and LLMs to securely connect to Hister. This means you can ask an AI, "What did I read about PostgreSQL optimization last week?" and it can query your private Hister index to find the answer, dramatically enhancing personal AI context.