# Opyt Opyt turns the bookmarks, follows, likes, and subscriptions you already make into a growing knowledge base for your AI client. It serves that source material through MCP so clients such as Claude, Cursor, Codex, and Windsurf can search, read, and analyze it. - Findings24: - Website: - Pricing: Freemium - Categories: AI Knowledge Base, AI Research Assistant, MCP Tools ## What is Opyt? Opyt is a personal knowledge-base and research layer built from the sources a user already follows, saves, likes, or subscribes to. It imports material from selected sources, indexes the full text where supported, and serves the resulting store to an AI client through the Model Context Protocol (MCP). The product is intended for AI-assisted research grounded in a user’s own curation. Users can search across saved and tracked material, read a topic in publication order, inspect claims and changes in an author’s position, discover new work related to standing questions, and draft answers with citations to source URLs. When the store lacks evidence, Opyt can expose those gaps rather than silently filling them with general web knowledge. A local installation works with MCP-compatible clients such as Claude Code, Claude Desktop, Cursor, Codex, and Windsurf. The site also describes a hosted connector for Claude.ai. Opyt’s documented source coverage includes X, Substack, GitHub, public blogs, arXiv, OpenAlex, LinkedIn content marked as coming soon, and individual supported links. ## What can Opyt do? ### Curated source ingestion Opyt uses existing bookmarks, likes, follows, subscriptions, and saved posts as input. Users begin by selecting and confirming a source before Opyt reads it. ### Expanded author and archive tracking After a person is confirmed, Opyt can find and verify additional public platforms and retrieve the person’s broader archive rather than limiting results to the items already saved. ### Cross-source search Search can combine posts, papers, repositories, and other indexed material in one ranked result list, with source links and indications of why a result is relevant to the user’s curation. ### Topic reading and analysis Reading workflows process documents in publication order and can return claims, briefings, trajectories, support and opposition, author changes of position, and questions that remain unresolved. ### Question-driven discovery Opyt turns topics into ongoing questions and checks sources such as arXiv, GitHub, and OpenAlex for new work. New items can be staged and ranked by how many of the user’s questions reached them. ### MCP delivery to AI clients A local Opyt installation exposes the store through MCP, allowing compatible AI clients to reason over the collected material. The site also describes a hosted connector for Claude.ai. ## Use Cases ### Check an idea before building Search across saved posts, papers, and repositories before committing to a project. This can reveal existing implementations, relevant criticism, and gaps in the material the user has already chosen to follow. ### Understand how a debate changed Read a topic chronologically and request a claims or trajectory view to distinguish enduring disagreement from an author or group changing position over time. ### Catch up after time away Ask what arrived against existing standing questions. Opyt can stage recent papers, repositories, and posts and rank them by the number of independent questions that reached each item. ### Pressure-test a decision Search the user’s selected sources on a business or technical decision, compare supporting and opposing views, and open the exact source behind a conclusion. ### Write from a bounded source set Ask an AI client to draft an argument using only material in the Opyt store. The documented workflow includes live source URLs and visible gaps where the collection does not support a claim. ## Frequently Asked Questions ### How does Opyt get content into its knowledge base? It uses sources a user already engages with, including X bookmarks, likes, follows and lists; Substack follows, subscriptions and saved posts; tracked GitHub repositories; public blogs; selected arXiv and OpenAlex research; and supported links supplied by the user. The user chooses and confirms a source before Opyt reads it. ### Which AI clients can use Opyt? The site describes a local MCP server that can be used by MCP-compatible clients including Claude Code, Claude Desktop, Cursor, Codex, and Windsurf. It also describes a hosted connector for Claude.ai. ### What kind of output does Opyt provide? Its documented workflows include ranked search results, chronological topic reads, claims analyses, briefings, trajectories, author-position changes, ongoing questions, and source-grounded drafts. Results can include source URLs, and missing evidence can be marked as a gap. ### Does Opyt automatically keep topics up to date? The site says Opyt continues running questions generated from a user’s topics against sources such as arXiv, GitHub, and OpenAlex, so new work can be available when the user returns without manually refreshing the topic. ### How much does Opyt cost? No pricing can be confirmed from the supplied site evidence. The listed pricing URL is unavailable, so users should check the product site for current availability and terms. ## Quick Facts - **Category**: AI research and personal knowledge base - **Primary workflow**: User-curated sources served to AI clients over MCP - **Local clients**: Claude Code, Claude Desktop, Cursor, Codex, and Windsurf - **Documented sources**: X, Substack, GitHub, public blogs, arXiv, OpenAlex, and supported links - **Hosted access**: Hosted connector described for Claude.ai - **Pricing**: Not confirmed; the listed pricing page is unavailable