Chavaphon Intarasuriyawong@bardlightyear

memory-dev is a local-first CLI tool that gives you AI-powered semantic memory over any git repository. You run memory-dev init on a repo and it walks every file, parses code at the AST level using Tree-sitter (extracting typed chunks like functions, classes, and modules), generates vector embeddings, and stores everything in a local SQLite database via Prisma. No cloud database needed — all data stays on your machine. Once indexed, memory-dev ask "..." embeds your natural-language question, finds the most relevant code chunks via cosine similarity, and streams back a synthesized answer with cited source locations. It also ships a watch mode for live re-indexing, a search command for raw vector lookup, and a status command for repo health. On top of the CLI, running memory-dev with no arguments launches an Ink TUI dashboard in the terminal. There's also an MCP server (mcp-server.ts) that exposes the same query capabilities as tools Claude Code and similar agents can call directly. The only external dependency is a Google AI API key (free tier) for generating embeddings and answers.

Build solutions that improve how software is built. Work with an AI that understands your codebase, helping you reduce repetitive work and build with real context. ⏳ Build your project in 48 hours. 🚀 Get hands-on with IBM Bob and explore how AI fits into real development workflows before it becomes standard. 💡 Build tools and workflows that developers would actually use. ⭐ Learn directly from experts throughout the event. 🤝 Join a team or build solo. 💰 Prize pool: $10,000 🧑🏻‍💻 Register before the Kick-Off Stream and start building with Bob.