How ashvanth-kumar/codequeryx works
CodeQuery is a Streamlit app for asking questions about public GitHub repositories. Its main workflow is: load a repository, clone and parse its code, split files into chunks, build a vector index, retrieve relevant chunks for a question, and generate an answer with Groq. The app also displays source previews and tracks chat history. README capabilities are represented, including Sentence Transformers embeddings and the documented FAISS store. The graph follows calls shown in app.py and module excerpts; unsampled storage details and any other integrations are not inferred.
Main entry points: app.py Streamlit chat app · repo_loader.py Repository loader · rag_engine.py Context and answer engine · vector_store.py Embedding and vector search
Components 11
User interface
- Streamlit chat app
app.py - Answer and sources
app.py
Repository ingestion
- Repository loader
repo_loader.py - Clone repository
repo_loader.py - Parse code files
repo_loader.py - Chunk code
chunker.py
Search and answers
- Embedding and vector search
vector_store.py - Context and answer engine
rag_engine.py
Other components
- Questioner
- GitHub repository
- Groq LLM API
Connections 11
- Questioner → Streamlit chat app: submits URL and questions
README.md - Streamlit chat app → Repository loader: loads repository
app.py - GitHub repository → Clone repository: clones from
README.md - Repository loader → Clone repository: calls clone_repo
repo_loader.py - Repository loader → Parse code files: calls parse_files
repo_loader.py - Streamlit chat app → Chunk code: chunks documents
app.py - Streamlit chat app → Embedding and vector search: builds index
app.py - Streamlit chat app → Embedding and vector search: searches query
app.py - Streamlit chat app → Context and answer engine: answers with results
app.py - Context and answer engine → Groq LLM API: requests completion
rag_engine.py - Streamlit chat app → Answer and sources: shows answer and sources
app.py