Configuration

While codectx works out of the box with zero configuration, you can fine-tune its behavior.

The .codectx.toml file

Place a .codectx.toml file in the root of your repository (or add a [tool.codectx] section to pyproject.toml) to persist configuration options.

.codectx.toml

# Explicit file patterns to ignore
extra_ignore = [
    "**/generated/**",
    "*.draft.py"
]

# The token constraint limit for generated Context output
token_budget = 100000

# Destination relative to cwd
output_file = "CONTEXT.md"

# Generate multilayer output files alongside the main file
layers = false

# Enable/disable git tracking algorithms
no_git = false

# Use language model embeddings or skip over heuristics based
llm_enabled = false
llm_provider = "openai"
llm_model = ""

# Track the project since specific interval string
since = "7 days ago"

CLI Overrides

Any value in the .codectx.toml file can be overridden via the CLI during generation:

codectx analyze . --tokens 50000 --no-git

Ignoring Files

By default, codectx ignores binaries, cache, and items within your standard .gitignore and an optional .ctxignore file. You can augment this using extra_ignore arrays within .codectx.toml or the pyproject.toml tool key.