Hey HN, we’re the team at Morph Labs and we’re excited to release Phorm (<a href="https://phorm.ai" rel="nofollow">https://phorm.ai</a>), a fast, simple, and SOTA codebase answer engine. You can search over up to 8 repositories in almost any language, and Phorm can comfortably handle repositories up to ~200K LOC each. It is free during our initial research preview.<p>Phorm’s Advanced Indexing combines synthetic data with static analysis of the code graph to improve the relevancy of search results by up to 3X. We’re proud to launch with featured Advanced Indexing support for a select group of leading open-source projects:<p>- Nomic AI (<a href="https://t.ly/vDIn0" rel="nofollow">https://t.ly/vDIn0</a>)
- LlamaIndex (<a href="https://t.ly/xxsSP" rel="nofollow">https://t.ly/xxsSP</a>)
- Unstructured.io (<a href="https://t.ly/08Vi8" rel="nofollow">https://t.ly/08Vi8</a>)
- Charm(<a href="https://t.ly/0KGzf" rel="nofollow">https://t.ly/0KGzf</a>)
- Turso (<a href="https://t.ly/qH5yu" rel="nofollow">https://t.ly/qH5yu</a>), and
- Axolotl (<a href="https://t.ly/FwqqT" rel="nofollow">https://t.ly/FwqqT</a>).<p>A bit of history: a few months ago we built <a href="https://moogle.ai" rel="nofollow">https://moogle.ai</a>, a specialized semantic search engine for the Lean mathematical library. Moogle is used by mathematicians like Terence Tao for formalizing their arguments in the Lean proof assistant. We saw how much users loved Moogle and realized that codebase-specific search could provide a lot of value. We realized we wanted a Moogle for everything, so we built one ourselves. Combining semantic search with our work on synthetic data generation for codebases led us to Phorm.<p>We hope you find it useful!