Show HN: I built ThreadShelf to reuse hard-to-export AI chats like OpenRouter

Category: ai-ml

Tags: local-first, semantic-search, ai-chat, mcp, vector-database

Score: 6.8/10 (Innovation: 6, Technical: 7, Documentation: 8, Utility: 6)

ThreadShelf is a local-first archive and semantic search tool that normalizes AI chat exports from six providers (ChatGPT, Claude, Google AI Studio, OpenRouter, LM Studio, Grok) into a unified LanceDB index, searchable via web UI, HTTP API, or MCP. It also supports continuing conversations through local llama.cpp inference or optional OpenRouter. It's interesting because it tackles the fragmented, hard-to-export nature of AI chat histories with a privacy-respecting, multi-frontend archive that doubles as a retrieval layer for LLM tooling.

Target audience: AI power users, developers, and researchers who want private, searchable archives of LLM conversations and MCP-compatible retrieval for agents

Repository: https://github.com/ChrystianSchutz/ThreadShelf · TypeScript · MIT · 6 stars

View on Hacker News