Configuration

All configuration is done through environment variables (.env in the project root, or your process manager / docker compose environment).

Database

Variable	Default	Description
DATABASE_URL	postgresql://aeoscope:aeoscope@localhost:5432/aeoscope	Postgres connection string

The app applies its migrations automatically on startup.

Engine API keys (BYOK)

Each hosted engine is enabled by setting its key; engines without a key are shown as unavailable in Settings.

Variable	Enables
OPENAI_API_KEY	OpenAI (ChatGPT)
PERPLEXITY_API_KEY	Perplexity
GEMINI_API_KEY	Google Gemini
ANTHROPIC_API_KEY	Anthropic (Claude)

The mock engine needs no key.

Model overrides

Every engine has a sensible default model. Override per engine:

Variable	Default
OPENAI_MODEL	gpt-4o-mini
PERPLEXITY_MODEL	sonar
GEMINI_MODEL	gemini-3.8-flash
GEMINI_SEARCH	off - set true to ground Gemini in live Google Search (adds source citations, billed by Google beyond its free tier)
ANTHROPIC_MODEL	claude-sonnet-4-5
OLLAMA_MODEL	llama3.2
CUSTOM_MODEL	local-model

Local servers

Variable	Default	Description
OLLAMA_BASE_URL	http://127.0.0.1:11434	Ollama address
CUSTOM_BASE_URL	-	OpenAI-compatible base URL, e.g. http://localhost:1234/v1 for LM Studio. Setting it enables the Custom engine. When aeoscope runs in Docker and the server runs on your machine, use http://host.docker.internal:<port>/v1 instead of localhost
CUSTOM_API_KEY	-	Only needed if your server requires auth

See Local models for per-tool examples, model identifiers, and the Docker notes.

Scheduler and cron endpoint

Variable	Default	Description
CRON_SECRET	-	Protects POST /api/cron. The endpoint is disabled when unset
AEOSCOPE_SCHEDULER	on	Set to off to disable the built-in scheduler

See Scheduling.