Configuration Files
The public Docker Compose deployment needs exactly two editable files:
.env.dockerconfig.yml.local
.env.docker
Create it from .env.compose.template:
bash
cp .env.compose.template .env.docker
The most important keys are:
| Key | Required | Purpose |
|---|---|---|
BUILD_MODE |
Yes | Must remain hub for the published Docker Hub flow |
DOCKER_REGISTRY |
Usually | Defaults to tylconsulting |
IMAGE_VERSION |
Usually | Defaults to latest; pin it for repeatable deployments |
OPENAI_API_KEY / ANTHROPIC_API_KEY / GEMINI_API_KEY / OPENROUTER_API_KEY |
At least one | LLM provider credentials |
EMBEDDING_PROVIDER |
Yes | Embedding backend used for vector retrieval |
EMBEDDING_MODEL |
Yes | Embedding model name |
EMBEDDING_API_KEY |
Usually | Key for the embedding backend |
BACKEND_AI_PROVIDER |
Yes | Backend provider used by Django tasks |
BACKEND_AI_MODEL |
Yes | Backend model identifier |
DJANGO_SUPERUSER_USERNAME |
Yes | Admin user bootstrap |
DJANGO_SUPERUSER_PASSWORD |
Yes | Admin password bootstrap |
WEB_PORT |
Yes | Main entrypoint exposed by the proxy |
FRONTEND_PORT |
Yes | Direct frontend port if needed |
SQL_GENERATOR_PORT |
Yes | Direct SQL Generator port if needed |
POSTGRES_INTERNAL |
Optional | Keep true for the default Compose setup |
config.yml.local
Create it from the tracked template:
bash
cp config.yml config.yml.local
This file controls:
- enabled AI providers
- embedding configuration
- backend default model
- supported database drivers
- admin and demo bootstrap users
- ports and runtime logging defaults
- relevance guard thresholds used by the SQL Generator
Minimum Recommended Edits
Before the first boot, update:
- at least one provider in
ai_providers embedding.provider,embedding.model, and its API keybackend_ai_modeladmin.password- any host port overrides if
8040,3040, or8020are already occupied
Sensitive Files
Do not commit:
.env.dockerconfig.yml.localif it contains real credentials
Keep the tracked files as templates only.