FCodeAgent

Configuration

Step 1: Update Paths

Edit the following paths in app.py to match your environment:

FUNCTIONS_ROOT = Path("Your Functions Path")         # Path to function datasets
GRAPHS_ROOT = Path("Your Graphs Path")              # Path to dependency graphs
COMMENT_LIBRARY_ROOT = Path("Your Comment Library Path")  # Path to comment libraries

Step 2: Update Generation & Agent Configuration

Modify config.py to adjust model and generation parameters:

# Example: Change model to gpt-4o-mini
LLM_CONFIG["model"] = "gpt-4o-mini"

# Adjust temperature for specific agents
LLM_CONFIG["temperature"]["generator"] = 0
LLM_CONFIG["temperature"]["decomposer"] = 0

# Modify generation settings
GENERATION_CONFIG["include_docstrings"] = True
GENERATION_CONFIG["include_type_hints"] = True
GENERATION_CONFIG["coding_style"] = "pep8"

# Modify framework parameters
FRAMEWORK_CONFIG["max_retries"] = 5

Step 3: Update LLM API Base and API Key

Modify llm_client.py to configure your large model API credentials.
For example, for OpenAI, Qwen, Gemini, or DeepSeek clients:

Step 4: Run the Server

Start the backend service by running:

python app.py

Default host: 0.0.0.0 Default port: 8081

Once the server is running, you can access:

Home: http://:/ — displays server status. Swagger UI: http://:/swagger — view API documentation and test endpoints. Frontend interface: connect via SocketIO to interactively call backend APIs.