Show HN: A Local-First Agentic Knowledge Manager (github.com)

Kept has launched a local-first knowledge manager for AI conversations, allowing users to store their interactions with models like ChatGPT, Claude, and others as Markdown files on their machines. This innovative tool enables users to search, browse, and connect their chat histories via a desktop application, which creates an Obsidian-compatible vault and supports local indexing for enhanced retrieval of past discussions. By keeping user data decentralized, Kept empowers individuals with full ownership of their chat logs while sidestepping the limitations of vendor UIs. The significance of Kept lies in its ability to transform fleeting AI discussions into structured, accessible knowledge repositories. Key technical features include fast local search powered by SQLite FTS5, support for multiple chat providers, and a local server enabling interaction with AI models directly over user archives. With built-in tools for topic discovery, project management, and a graph view to visualize connections, Kept represents a substantial advancement for developers and researchers looking to leverage their AI dialogues for ongoing learning and idea generation. The tool's commitment to user privacy, maintaining all data locally without cloud dependencies, highlights its potential impact on how individuals manage and utilize AI-generated content.