aksharpatel.dev
AI · Codebase · Explainer

Understand any codebase,
instantly.

Paste a GitHub URL — we'll clone, chunk, embed, and let you ask natural-language questions about the code, grounded in real source citations.

Source code is sent to OpenAI's API for embedding and analysis. Only submit public repos you're comfortable sharing.

Public repos only≤ 50 MB≤ 1,000 files≤ 5,000 chunks.py .js .ts .go +more

Deep chunking

Source files split into 1,500-char semantic segments, fully indexed.

Vector search

pgvector similarity search retrieves the top-5 most relevant chunks.

Grounded answers

Every answer cites the exact file and line range it was drawn from.

A personal RAG pipeline — OpenAI embeddings + pgvector, grounded in real source code.

Personal · Demo
FastAPINext.js 15PostgreSQLpgvectorSQLAlchemyAlembicOpenAI Embeddingsgpt-4o-miniDockerAWS App RunnerVercel