Opyt

Product Introduction

1. Definition: Opyt is an open-source, MIT-licensed MCP (Model Context Protocol) server that functions as an automated, self-curating knowledge base. It is a technical tool that aggregates, indexes, and semantically analyzes content from a user's existing digital footprint.
2. Core Value Proposition: Opyt exists to transform passive information consumption into an active, structured, and growing private knowledge graph. Its primary value is automating the curation and contextualization of information from disparate sources (like X, Substack, GitHub, arXiv) into a single, queryable store that evolves autonomously through AI-driven discovery.

Main Features

1. Automated Source Aggregation: Opyt connects to platforms where users already engage (X bookmarks/likes/follows, Substack subscriptions, GitHub repos, personal blogs, arXiv) and ingests their full archives. It uses managed browser sessions (for private platforms like X) and public APIs/feeds (for GitHub, arXiv) to pull content in full text, creating a local SQLite database or a hosted copy.
2. AI-Powered Topic Synthesis & Autonomous Discovery: The system's core innovation is the sitting tool. It doesn't just store data; it reads through aggregated content on a topic, uses an LLM (via OpenRouter) to generate its own follow-up questions, and continuously runs these questions against research databases (arXiv, OpenAlex) and code repositories (GitHub) to discover and ingest new, relevant work without user intervention.
3. MCP-Native Retrieval & Analysis Tools: As an MCP server, Opyt exposes its capabilities as tools within compatible AI clients (Claude Desktop, Cursor, Windsurf). Key tools include search for semantic retrieval, sitting with specialized lenses ("claims," "briefing," "trajectory," "disconfirmation") for deep analytical reads, and aggregate for topic trend analysis. This eliminates the need for a separate chat UI, integrating directly into the developer's existing workflow.
4. Frictionless Knowledge Sharing: The share and accept tools allow users to publish their entire knowledge base to a hosted Opyt instance. Others can then query this shared KB directly from their client, enabling collaborative research where one user's curated insights become a searchable resource for others, with access controlled by the sharer.

Problems Solved

1. Pain Point: Information fragmentation and context loss. Critical insights are siloed across dozens of platforms (bookmarks, subscription emails, paper PDFs, starred repos), making it impossible to search, cross-reference, or gain a synthesized view of one's own accumulated knowledge.
2. Target Audience: The primary user personas are Knowledge Workers and Researchers (analysts, academics, strategists), Technology Professionals (software engineers, data scientists, CTOs) who follow fast-moving fields, and Curious Generalists who actively follow diverse thinkers and topics across multiple platforms.
3. Use Cases: Essential for deep literature reviews where tracking evolving arguments is key; for competitive intelligence by automatically tracking key individuals and projects; for maintaining a personal, evergreen reference library that grows smarter on its own; and for team-based research where sharing foundational reading eliminates duplicate work.

Unique Advantages

1. Differentiation: Unlike traditional bookmark managers (Pocket, Raindrop.io) or read-later apps, Opyt performs deep, semantic ingestion and autonomous discovery. Unlike manual note-taking apps (Obsidian, Notion), it automates the collection and initial synthesis. Unlike AI search tools (Perplexity), it builds on a persistent, private corpus you have curated, providing personalized, citation-grounded answers.
2. Key Innovation: The "self-questioning" autonomous discovery loop. The system's ability to read a topic, generate its own research questions using LLMs, and use those questions as standing queries to find new material is a unique approach to continuous knowledge base growth. This turns the knowledge base from a static archive into an active research assistant.

Frequently Asked Questions (FAQ)

1. Is Opyt free and open source? Yes, Opyt is completely free to use and is released under the permissive MIT open-source license. Costs are incurred only for the LLM calls via OpenRouter for processing and embedding tasks.
2. Where is my data stored with Opyt? You have two options. For full privacy, you can run Opyt locally where all data resides in a SQLite database (~/.opyt) on your machine. Alternatively, you can use the hosted connector for Claude.ai or ChatGPT, where your data is stored on Opyt's servers, accessible via a Google OAuth sign-in.
3. How does Opyt access my private X or Substack data? For private platforms, Opyt uses a managed browser session. You authenticate once within the tool's flow, and it maintains a session to access only the data you have explicit access to (your bookmarks, your follows, your subscriptions). It does not store your platform passwords.
4. Can I remove data from my Opyt knowledge base? Yes. The forget tool allows you to delete specific items or stop tracking a person entirely. For synced sources (like X bookmarks), an item will be re-added if it still exists in the original source (e.g., you didn't un-bookmark it), giving you control at the source.
5. What AI models does Opyt use? Opyt uses OpenRouter as its LLM gateway, allowing you to choose and pay for any model supported by that platform (like Claude, GPT, or open-source models) for its classification, extraction, and embedding tasks. Your primary AI client (Claude, ChatGPT, etc.) handles the final reasoning and chat interaction.