Configuration

No configuration required — Context42 works out of the box. All settings below are optional and only needed if you want to customize the default behavior.

Environment variables

Variable	Required	Description	Default
`C42_DATA_DIR`	Optional	Directory for storing index data	`~/.local/share/context42`
`C42_LOG_LEVEL`	Optional	Logging level (`DEBUG`, `INFO`, `WARNING`, `ERROR`)	`INFO`
`C42_EMBEDDING_MODEL`	Optional	Sentence-transformers model for embeddings	`BAAI/bge-small-en-v1.5`
`C42_CHUNK_SIZE`	Optional	Characters per chunk when splitting documents	`500`
`C42_BATCH_SIZE`	Optional	Chunks processed per batch during indexing	`50`
`HF_TOKEN`	Optional	Hugging Face token for faster model downloads	—

Embedding model

By default Context42 uses BAAI/bge-small-en-v1.5, a lightweight model that runs efficiently on CPU without requiring a GPU. The model is downloaded automatically on the first indexing run.

To use a different sentence-transformers model:

export C42_EMBEDDING_MODEL="BAAI/bge-base-en-v1.5"

Tip: If you have a GPU available, larger models like BAAI/bge-large-en-v1.5 can provide better search quality at the cost of higher resource usage.

After changing the embedding model you must re-index your sources:

c42 index

Hugging Face token

Set the HF_TOKEN environment variable to enable faster model downloads and avoid rate limiting:

export HF_TOKEN="hf_your_token_here"

Get your token at huggingface.co/settings/tokens.

Data directory

Context42 stores its vector index and metadata in the data directory. By default this is:

Platform	Path
macOS	`~/Library/Application Support/context42`
Linux	`~/.local/share/context42`
Windows	`%LOCALAPPDATA%\context42`

Override with the C42_DATA_DIR environment variable:

export C42_DATA_DIR=/path/to/custom/dir

Supported file formats

Context42 indexes the following file types:

Markdown (.md) — Full support with heading-aware chunking
reStructuredText (.rst) — Full support

More formats are planned for future releases.