Installation
Prebuilt Binary
curl -sSL https://raw.githubusercontent.com/idelchi/aura/refs/heads/dev/install.sh | sh -s -- -d ~/.local/bin
From Source
go install github.com/idelchi/aura@latest
Provider Setup
Aura requires at least one LLM provider. Configure providers in .aura/config/providers/.
Supported providers:
| Provider | Type | Default URL | URL Override |
|---|---|---|---|
| Ollama | Local | None (must be set in YAML) | Yes |
| llama.cpp | Local | None (must be set in YAML) | Yes |
| OpenRouter | Cloud | https://openrouter.ai/api/v1 | Yes |
| OpenAI | Cloud | https://api.openai.com/v1 | Yes |
| Anthropic | Cloud | https://api.anthropic.com | Yes |
| Google Gemini | Cloud | https://generativelanguage.googleapis.com/ | Yes |
| GitHub Copilot | Cloud | Dynamic (token exchange with GitHub) | No |
| OpenAI Plus (Codex) | Cloud | https://chatgpt.com/backend-api/codex | Yes |
Copilot does not accept URL overrides — its URL is resolved dynamically via GitHub’s token exchange API.
See Provider Configuration for YAML setup and token configuration.
Initialize Configuration
Scaffold the default configuration directory:
aura init
This creates .aura/config/ with all default agents, modes, providers, features, and prompts. See Configuration for details.
Environment Variables
All CLI flags can be set via AURA_ prefixed environment variables. Precedence: CLI flag > AURA_* env var > default value.
# Examples
export AURA_AGENT=high
export AURA_MODEL=llama3:8b
export AURA_DEBUG=1
# See all resolved settings
aura --print-env
Tokens are resolved as AURA_PROVIDERS_<PROVIDER>_TOKEN. As an example, defining a provider ollama would look for AURA_PROVIDERS_OLLAMA_TOKEN. Tokens can also be set in provider YAML configs or loaded from an env file via --env-file (default: secrets.env).