Analyze, audit and document your code in seconds — powered by AI.
Upload a .txt file or point at a public GitHub repository and get instant, structured feedback — from recruitment-grade audits to senior-level code reviews and auto-generated merge request descriptions.
This app is also deployed publicly on Streamlit Cloud: https://dev-tooling-chat.streamlit.app/
| Feature | Description |
|---|---|
| 🔍 Audit & Diagnostic | Evaluate a GitHub repo against a 10-point professional audit grid used by technical recruiters. |
| 📝 Senior Code Review | Get a comprehensive Python code review with weighted scoring, strengths, weaknesses and actionable recommendations. |
| 🔀 Merge Request Description | Auto-generate a complete, structured MR description from a git diff between two branches. |
| 🔄 Reset Context | Instantly reset the analysis state with a single click to start a fresh audit without reloading the page. |
- ✨ Modern UI — built with a custom dark theme, glassmorphism effects, and smooth animations.
- 🛡️ Robust Ingestion — automatically filters large files (
.lock, images, etc.) to optimize token usage and avoid rate limits. - ⚡ Smart Caching — reuses cloned repositories and analysis context to speed up subsequent requests.
Every feature supports two input methods:
- 📄 File upload — paste or export your code / diff as a
.txtfile. - 🔗 GitHub URL — provide a public repository URL; the app clones the repo and processes it automatically (via gitingest).
All AI responses can be copied to clipboard or downloaded as .txt with one click.
dev_tooling_chat/
├── app.py # Streamlit entry point & page routing
├── views/
│ ├── __init__.py
│ ├── audit_diagnostic.py # 🔍 Audit & Diagnostic page
│ ├── code_review.py # 📝 Senior Code Review page
│ └── merge_request.py # 🔀 Merge Request Description page
├── src/
│ └── dev_tooling_chat/
│ ├── __init__.py
│ ├── styles.py # Global CSS design system (dark theme)
│ └── utils.py # Shared utilities (LLM, Git, UI helpers)
├── prompts/
│ ├── repo_recrutement.txt # System prompt for Audit & Diagnostic
│ ├── code_audit.txt # System prompt for Code Review
│ └── mr_assistant.txt # System prompt for MR Description
├── .streamlit/
│ └── config.toml # Streamlit theme configuration
├── Makefile # Common dev commands
├── pyproject.toml # Project metadata & dependencies
├── uv.lock # Lockfile (uv)
├── Dockerfile # Docker image definition
└── .dockerignore # Docker build exclusions
| Requirement | Version |
|---|---|
| Python | ≥ 3.12 |
| uv | latest recommended |
| Git | any recent version |
| A Groq API key | free tier is fine |
# Clone the repository
git clone https://github.com/<your-username>/dev_tooling_chat.git
cd dev_tooling_chat
# Create a virtual environment and install dependencies
uv syncuv run streamlit run app.pyThe application will open in your browser at http://localhost:8501.
| Command | Description |
|---|---|
make install |
Install all dependencies via uv sync |
make run |
Start the Streamlit app |
make run-dev |
Start with auto-reload & debug logging |
make lint |
Run the ruff linter |
make lint-fix |
Run ruff with auto-fix |
make format |
Format code with ruff |
make check |
Run all code quality checks (lint + format) |
make typecheck |
Run mypy type checking |
make build |
Build the distribution package |
make clean |
Remove caches and build artifacts |
make help |
Show all available targets |
You can also run the application using Docker:
# Build the image
docker build -t dev-tooling-chat .
# Run the container
# Replace <your_groq_api_key> with your actual API key
docker run -p 8501:8501 -e GROQ_API_KEY=<your_groq_api_key> dev-tooling-chat- Enter your Groq API key in the sidebar (starts with
gsk_...). - Select an AI model from the dropdown — only text-generation models are listed.
- Pick a feature from the sidebar navigation:
- Audit & Diagnostic — upload code or provide a GitHub URL to run a recruitment audit.
- Senior Code Review — upload code or provide a GitHub URL for a detailed code quality review.
- Merge Request Description — upload a diff file, or provide a GitHub URL, select source/target branches, and generate a structured MR description.
- Review the AI response, then use the 📋 Copy or 📥 Download buttons.
┌──────────────────────────────────────┐
│ Streamlit UI │
│ (app.py + views/ + styles) │
├──────────────────────────────────────┤
│ dev_tooling_chat.utils │
│ ┌───────────┬──────────┬──────────┐ │
│ │ LLM layer │ Git │ UI │ │
│ │ (Groq API)│ helpers │ helpers │ │
│ └───────────┴──────────┴──────────┘ │
├──────────────────────────────────────┤
│ prompts/*.txt │
│ (system prompts per feature) │
└──────────────────────────────────────┘
app.py— Configures the Streamlit page, renders the sidebar (API key input, model selector, navigation), and routes to the appropriate view.views/— Each feature is a self-contained module exposing arender()function.src/dev_tooling_chat/styles.py— Global CSS design system providing the dark theme, glassmorphism effects, gradient accents, and responsive styling.src/dev_tooling_chat/utils.py— Shared logic split into three sections:- LLM —
fetch_groq_models()andcall_groq_llm()for interacting with the Groq chat API. - Git helpers —
clone_repo(),clone_and_ingest(),get_branches(),git_diff()for repository operations. - UI helpers —
load_prompt()andrender_response_actions()(copy + download buttons).
- LLM —
prompts/— Plain-text system prompts injected into the LLM call for each feature.
| Package | Purpose |
|---|---|
| streamlit ≥ 1.44 | Web UI framework |
| groq ≥ 0.15 | Groq API Python client |
| gitpython ≥ 3.1 | Git operations (clone, diff, branches) |
| requests ≥ 2.31 | HTTP calls to the Groq models endpoint |
| loguru ≥ 0.7 | Structured logging |
Runtime tool: The Audit & Diagnostic and Code Review features also invoke
gitingestviauv run --with gitingest gitingest .to digest a cloned repository into a single text file.
- Fork the repository.
- Create a feature branch:
git checkout -b feat/my-feature. - Commit your changes:
git commit -m "feat: add my feature". - Push to the branch:
git push origin feat/my-feature. - Open a Pull Request.
This project is licensed under the MIT License.