The primary entrypoint to programs leveraging Codegen is the Codebase class.

​
Local Codebases

Construct a Codebase by passing in a path to a local git repository or any subfolder within it. The path must be within a git repository (i.e., somewhere in the parent directory tree must contain a .git folder). 
from codegen import Codebase
from codegen.sdk.enums import ProgrammingLanguage

# Parse from a git repository root
codebase = Codebase("path/to/repository")

# Parse from a subfolder within a git repository
codebase = Codebase("path/to/repository/src/subfolder")

# Parse from current directory (must be within a git repo)
codebase = Codebase("./")

# Specify programming language (instead of inferring from file extensions)
codebase = Codebase("./", programming_language=ProgrammingLanguage.TYPESCRIPT)
By default, Codegen will automatically infer the programming language of the codebase and parse all files in the codebase. You can override this by passing the programming_language parameter with a value from the ProgrammingLanguage enum.
The initial parse may take a few minutes for large codebases. This pre-computation enables constant-time operations afterward. Learn more here.

​
Remote Repositories

To fetch and parse a repository directly from GitHub, use the from_repo function. 
import codegen
from codegen.sdk.enums import ProgrammingLanguage

# Fetch and parse a repository (defaults to /tmp/codegen/{repo_name})
codebase = codegen.from_repo('fastapi/fastapi')

# Customize temp directory, clone depth, specific commit, or programming language
codebase = codegen.from_repo(
    'fastapi/fastapi',
    tmp_dir='/custom/temp/dir',  # Optional: custom temp directory
    commit='786a8ada7ed0c7f9d8b04d49f24596865e4b7901',  # Optional: specific commit
    shallow=False,  # Optional: full clone instead of shallow
    programming_language=ProgrammingLanguage.PYTHON  # Optional: override language detection
)
Remote repositories are cloned to the /tmp/codegen/{repo_name} directory by default. The clone is shallow by default for better performance.

​
Configuration Options

You can customize the behavior of your Codebase instance by passing a CodebaseConfig object. This allows you to configure secrets (like API keys) and toggle specific features: 
from codegen import Codebase
from codegen.sdk.codebase.config import CodebaseConfig, GSFeatureFlags, Secrets

codebase = Codebase(
    "path/to/repository",
    config=CodebaseConfig(
        secrets=Secrets(
            openai_key="your-openai-key"  # For AI-powered features
        ),
        feature_flags=GSFeatureFlags(
            sync_enabled=True,  # Enable graph synchronization
            ...  # Add other feature flags as needed
        )
    )
)
The CodebaseConfig allows you to configure:
  • secrets: API keys and other sensitive information needed by the codebase
  • feature_flags: Toggle specific features like language engines, dependency management, and graph synchronization
For a complete list of available feature flags and configuration options, see the source code on GitHub.

​
Advanced Initialization

For more complex scenarios, Codegen supports an advanced initialization mode using ProjectConfig. This allows for fine-grained control over:
  • Repository configuration
  • Base path and subdirectory filtering
  • Multiple project configurations
Here’s an example: 
from codegen import Codebase
from codegen.git.repo_operator.local_repo_operator import LocalRepoOperator
from codegen.git.schemas.repo_config import BaseRepoConfig
from codegen.sdk.codebase.config import ProjectConfig
from codegen.sdk.enums import ProgrammingLanguage

codebase = Codebase(
    projects = [
        ProjectConfig(
            repo_operator=LocalRepoOperator(
                repo_path="/tmp/codegen-sdk",
                repo_config=BaseRepoConfig(),
                bot_commit=True
            ),
            programming_language=ProgrammingLanguage.TYPESCRIPT,
            base_path="src/codegen/sdk/typescript",
            subdirectories=["src/codegen/sdk/typescript"]
        )
    ]
)
For more details on advanced configuration options, see the source code on GitHub.

​
Supported Languages

Codegen currently supports: