Kodit

--- title: Configuration Reference description: Kodit Configuration Reference weight: 29 --- This document contains the complete configuration reference for Kodit. All configuration is done through environment variables. {% for model_name, model_info in models.items() %} {%- if model_name != "BaseSettings" and model_info.env_vars %} ## {{ model_name }} {{ model_info.description }} | Environment Variable | Type | Default | Description | |---------------------|------|---------|-------------| {%- for env_var in model_info.env_vars %} | `{{ env_var.name }}` | {{ env_var.type }} | `{{ env_var.default }}` | {{ env_var.description }} | {%- endfor %} {% endif %} {% endfor %} ## Applying Configuration There are two ways to apply configuration to Kodit: 1. A local `.env` file (e.g. `kodit --env-file .env serve`) 2. Environment variables (e.g. `DATA_DIR=/path/to/kodit/data kodit serve`) How you specify environment variables is dependent on your deployment mechanism. ### Docker Compose For example, in docker compose you can use the `environment` key: ```yaml services: kodit: environment: - DATA_DIR=/path/to/kodit/data ``` ### Kubernetes For example, in Kubernetes you can use the `env` key: ```yaml env: - name: DATA_DIR value: /path/to/kodit/data ``` ## Example Configurations ### Enrichment Endpoints Enrichment is typically the slowest part of the indexing process because it requires calling a remote LLM provider. Ideally you want to maximise the number of parallel tasks but all services have rate limits. Start low and increase over time. See the [configuration reference](/kodit/reference/configuration/index.md) for full details. The following is a selection of examples. #### Helix.ml Enrichment Endpoint Get your free API key from [Helix.ml](https://app.helix.ml/account). ```sh ENRICHMENT_ENDPOINT_BASE_URL=https://app.helix.ml/v1 ENRICHMENT_ENDPOINT_MODEL=hosted_vllm/Qwen/Qwen3-8B ENRICHMENT_ENDPOINT_NUM_PARALLEL_TASKS=1 ENRICHMENT_ENDPOINT_TIMEOUT=300 ENRICHMENT_ENDPOINT_API_KEY=hl-XXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXX ``` #### Local Ollama Enrichment Endpoint ```sh ENRICHMENT_ENDPOINT_BASE_URL=http://localhost:11434 ENRICHMENT_ENDPOINT_MODEL=ollama_chat/qwen3:1.7b ENRICHMENT_ENDPOINT_NUM_PARALLEL_TASKS=1 ENRICHMENT_ENDPOINT_EXTRA_PARAMS='{"think": false}' ENRICHMENT_ENDPOINT_TIMEOUT=300 ``` #### Azure OpenAI Enrichment Endpoint ```sh ENRICHMENT_ENDPOINT_BASE_URL=https://winderai-openai-test.openai.azure.com/ ENRICHMENT_ENDPOINT_MODEL=azure/gpt-4.1-nano # Must be in the format "azure/azure_deployment_name" ENRICHMENT_ENDPOINT_API_KEY=XXXX ENRICHMENT_ENDPOINT_NUM_PARALLEL_TASKS=5 # Azure defaults to 100K TPM ENRICHMENT_ENDPOINT_EXTRA_PARAMS={"api_version": "2024-12-01-preview"} ```