Kept is a local-first archive tool designed to capture and organize conversations from popular large language model (LLM) chat platforms. It addresses the need for users to retain, search, and manage their AI chat data by saving each conversation as a plain markdown file on the user's filesystem, with no cloud storage required. The tool supports ChatGPT, Claude, Gemini, Grok, and Kimi, allowing users to archive chats from multiple providers in a consistent format.
The platform offers several features to enhance the usability and accessibility of archived conversations. Users can trigger manual capture of chats by typing @kept inline, which brings up a palette to save the full conversation, recent messages, or specific branches. Kept also provides a Chrome extension that enables background auto-sync, writing new conversations to the local vault at configurable intervals and ensuring no duplicate entries through SHA-256 deduplication. Full-text search is powered by SQLite FTS5, allowing filtering by platform, date, or message count, and supports archives with tens of thousands of conversations. Semantic search is available via Smart Recall, returning relevant passages in response to plain-English queries and providing citations to the source conversation. Additional features include automatic topic clustering, knowledge graph extraction of entities and relationships, and weekly digests and summaries in markdown format.
Kept is cross-platform, with installation options for Linux, macOS, Windows, and a Chrome extension. kept, and is compatible with any tool that can read markdown, such as text editors, ripgrep, or Obsidian. Users can bring their own API keys for OpenAI, Anthropic, or OpenRouter, or run local models via Ollama. The tool also enables code execution, integration with Claude Code projects, and provides an MCP server (kept-vault-server on npm) for interoperability with MCP-aware clients. PDF and image files can be added to conversations, with extracted text incorporated into the context for further processing.
Kept is distributed under the MIT license and is available as a public release on GitHub, with no premium tier or waitlist. All features are included in the public version, and no data leaves the user's machine by default, ensuring full local control over archived conversations.
In the LLM eval & observability space, Kept takes a focused approach. Preserving, searching, and organizing AI chat conversations locally without relying on cloud storage. Kept is an open-source project aimed at developers and power users who use multiple LLM chat platforms. The project is open source (MIT). The product ships for the command line, macOS, Windows, and Linux, and it can be self-hosted.
It is developed by @kept (community-driven), and the product first shipped in 2026. The project is developed in the open on GitHub with 113 stars and 42 commits in the last 90 days. Across PulseGate's embedding index, Kept has few near neighbours, marking it as relatively distinct. Among its 10 catalogued features are markdown archive, full-text search, and semantic recall. It exposes integrations via an MCP server.
Latest indexed changes and source events
Other apps tracked under the same category.
