Hey Product Hunt! 👋 Jumping into an unfamiliar codebase or taking over a collaborative group project always feels like trying to read a map written in a foreign language. Files are nested deep, documentation is outdated, and figuring out where authentication, proxy setups, or API routing lives eats up precious engineering hours. As a team building LemonBeam, we got tired of blindly pasting entire repositories into an AI assistant and hoping for the best. Instead of treating a codebase like a black box, we took a deterministic approach: LemonBeam classifies repository files, chunks code using Tree-sitter, and indexes metadata to generate accurate, source-backed architecture and contributor guides complete with citations. The Team Behind LemonBeam: Kanami Anderson architected the pipeline, led execution as Scrum Lead, and built the file discovery, classification stage, Tree-sitter chunkers, citation validator, and the local MCP server. John DeLong established the technical foundation, built the GitHub integration (repository validation, commit pinning, snapshot downloads), and spearheaded rigorous testing against OpenRouter and OpenAI. To fit into any developer's daily workflow, we built it across three modalities: The Web Experience: A clean React dashboard for visual exploration and team onboarding. The CLI: A terminal utility to generate local markdown guides without breaking your flow. The MCP Server: A local Model Context Protocol server that plugs directly into Claude Desktop or Claude Code for real-time architectural debugging and context. We would love for you to check out LemonBeam, test it on your favorite repository, and let us know what you think! Drop your feedback or questions below—we are hanging out in the comments all day. 👇

LemonBeam

AI starter guides from your JS/TS codebase structure

AI starter guides from your JS/TS codebase structure