DeepCodeGraph

DeepCodeGraph is an LLM agent that automatically generates UML and activity diagrams from codebases using PlantUML.

Installation

Install using pip:

pip install deepcodegraph

Installation from source:

git clone https://github.com/rhochgraf21/DeepCodeGraph/
pip install -e codegraph

Usage

The codegraph tool is designed to analyze your code repositories by generating dependency graphs or exporting the repository structure.

Below are the details on how to use the available commands and options.

Commands

• graph
  Visualize a repository’s dependency graph. Graphs are generated using the public PlantUML server.

• export
  Export the repository’s structure for further analysis of the agent progress.

Global Options

• -h, --help
  Show a help message and exit.

• -v, --verbose
  Increase verbosity. This flag can be used multiple times for more detailed output.

• --api-key API_KEY
  Specify your LLM API key. Alternatively, set the CODEGRAPH_API_KEY environment variable, or the API key environment variable for your provider as found in LiteLLM.

• --provider PROVIDER
  Choose the LLM provider. The default provider is gemini.

• --model MODEL
  Specify the model to use with the provider. The default is gemini-2.0-flash-exp.

• --path PATH
  Required if not using --github. Provide the local filesystem path to your repository.

• --github GITHUB_URL
  Required if not using --path. Provide the URL of the GitHub repository to analyze.

Example

To generate a dependency graph for a local repository located at /path/to/repo, run:

codegraph graph --path /path/to/repo

To analyze a GitHub repository, run:

codegraph graph --github https://github.com/username/repository

Run codegraph -h for more detailed information.

Note: You must provide either the --path option (for local repositories) or the --github option (for GitHub repositories). The GitHub functionality is provided to analyze repositories hosted on GitHub.

Supported Providers

DeepCodeGraph uses LiteLLM to provide access to models.

Provider	Supported Models (Examples)	Documentation Link
OpenAI	gpt-4o, o3-mini, etc.	OpenAI Docs
Anthropic	claude-3.7, claude-3.5, claude-3, etc.	Anthropic Docs
Azure	OpenAI models via Azure	Azure Docs
Google Vertex AI	Various models from Google	Google Vertex Docs
Google Gemini (AI Studio)	gemini-2.0-pro, gemini-2.0-flash, etc.	Google Gemini Docs
Mistral AI	mistral-small-latest, mistral-medium-latest, mixtral-8x7b, etc.	Mistral AI Docs
AWS Bedrock	Various models from Anthropic, Meta, Deepseek, Mistral, Amazon, etc.	AWS Bedrock Docs
OpenRouter	Various models available through OpenRouter	OpenRouter Docs
Huggingface	Various open-source models from Huggingface	Huggingface Docs
Cohere	command-r-plus, command-r, command, etc.	Cohere Docs
Ollama	Various models like mistral, gemma, llama3, etc.	Ollama Docs
Groq	llama3-8b, gemma-7b, mixtral-8x7b, etc.	Groq Docs

To connect to a provider, set CODEGRAPH_CUSTOM_API_ENV=1 and the LiteLLM environment variables specified in the docs, or set CODEGRAPH_API_KEY.

For example, for Gemini on OpenRouter:

# custom setup [api key variable inferred]
export CODEGRAPH_CUSTOM_API_ENV=1
export GEMINI_API_KEY=<your_api_key>

# explicit api key
export CODEGRAPH_API_KEY=<your_api_key>

Then set the CODEGRAPH_LLM_PROVIDER and CODEGRAPH_LLM_MODEL environment variables.

export CODEGRAPH_LLM_PROVIDER=openrouter
export CODEGRAPH_LLM_MODEL=google/gemini-2.0-flash-lite-preview-02-05:free

Note: The API key variable will be inferred from the provided model and provider unless CODEGRAPH_API_KEY is provided.

Contributing

Pull requests are welcome. For major changes, please open an issue first to discuss what you would like to change.

License

MIT