Environment Variables

Environment variables are runtime inputs loaded for the selected profile. EstaCoda reads them from the process environment and from the selected profile .env file.

The preferred storage location for secrets is:

~/.estacoda/profiles/<profile-id>/.env

Setup flows write secrets there with 0600 permissions. You can also reference an existing environment variable by name in config.

State isolation

Variable	Purpose
ESTACODA_HOME	Override the default state root (~/.estacoda). Use this to run dev builds against isolated state without touching your real user data.

LLM provider API keys

Variable	Provider	Status
KIMI_API_KEY	Kimi	Live-proven
OPENAI_API_KEY	OpenAI	Live-proven
DEEPSEEK_API_KEY	DeepSeek	Live-proven
OPENROUTER_API_KEY	OpenRouter	Live-proven
OPENAI_COMPATIBLE_API_KEY	Local / Custom OpenAI-compatible endpoint	Optional for local; default for custom OpenAI-compatible routes
GOOGLE_API_KEY	Google	Configurable/catalog-known
ANTHROPIC_API_KEY	Anthropic	Catalog-known; not runnable as a primary LLM route

MiniMax and Nous are catalog-known but not runnable in the current build.

Codex OAuth

Codex authentication stores tokens in ~/.estacoda/auth.json after OAuth device-code flow. The env secret store does not manage Codex tokens.

Voice provider keys

Variable	Purpose
VOICE_TOOLS_OPENAI_KEY	Default OpenAI audio key for TTS/STT
OPENAI_API_KEY	OpenAI audio fallback only when the configured OpenAI audio env is the default VOICE_TOOLS_OPENAI_KEY
ELEVENLABS_API_KEY	ElevenLabs TTS
MINIMAX_API_KEY	MiniMax TTS
GEMINI_API_KEY	Gemini TTS
XAI_API_KEY	xAI native TTS/STT
GROQ_API_KEY	Groq STT
HF_HOME	Optional faster-whisper / Hugging Face model cache root
TRANSFORMERS_CACHE	Optional Hugging Face cache env respected by the worker environment

Voice config stores environment-variable references only. Setup Editor and Onboarding Wizard collect hosted Voice provider API keys through masked input and write real key values to the selected profile .env only after reviewed apply. Raw Voice API keys are not stored in config, shown in review manifests, or inserted into prompt context. Runtime config defaults provide Voice models, voices, and provider settings, so interactive setup does not ask for model strings or env-var reference names.

No-key Voice providers are STT local and TTS edge. Edge does not require an API key, but it is networked: synthesis text is sent to Microsoft's Edge speech service and is not local/offline. Hosted key-required providers are STT openai, groq, and xai; and TTS openai, elevenlabs, minimax, gemini, and xai. STT mistral and TTS mistral, neutts, and kittentts are deferred/stub providers. Existing credentials can be reused through env-var references such as VOICE_TOOLS_OPENAI_KEY, OPENAI_API_KEY, GEMINI_API_KEY, GROQ_API_KEY, and XAI_API_KEY, and through existing compatible provider config/routes where supported. There are no voice credential pools, gateway brokers, managed fallbacks, or non-env sources.

Managed local STT defaults to ~/.estacoda/cache/huggingface for faster-whisper model cache when hfHome is not configured. If TRANSFORMERS_CACHE is already set in the process environment, runtime preserves it. The managed Python venv remains separate at ~/.estacoda/python-env.

Image generation keys

Variable	Provider
FAL_KEY	FAL
BYTEPLUS_ARK_API_KEY	BytePlus / Seedream
ARK_API_KEY	BytePlus / Seedream examples; reused by reviewed setup when already present
OPENAI_API_KEY	OpenAI / GPT Image

Browser provider keys

Variable	Provider	Purpose
BROWSERBASE_API_KEY	Browserbase	Browserbase API authentication.
BROWSERBASE_PROJECT_ID	Browserbase	Browserbase project used for cloud browser sessions.

These credentials satisfy Browserbase readiness only. They do not approve billable session creation. Browserbase sessions remain blocked until browser.cloudSpendApproved === true, normally set with estacoda browser approve-cloud and revoked with estacoda browser revoke-cloud.

Web search provider keys

Variable	Provider	Purpose
BRAVE_SEARCH_API_KEY	Brave Search	Default env var for Brave Search API authentication.

Brave config stores the env var reference at web.brave.apiKeyEnv; it does not store the raw key. DDGS does not use an API key. DDGS readiness comes from the managed Python capability:

estacoda python-env setup ddgs
estacoda python-env verify ddgs

Channel keys

Variable	Channel
ESTACODA_TELEGRAM_BOT_TOKEN	Telegram bot token used by guided setup
ESTACODA_DISCORD_TOKEN	Discord bot token
ESTACODA_WHATSAPP_BRIDGE_INSTALL_TIMEOUT	Optional timeout, in milliseconds, for the explicit estacoda whatsapp bridge dependency repair/install step

Email uses passwordEnv config keys that reference arbitrary env vars, for example EMAIL_PASSWORD.

WhatsApp does not use a bot-token environment variable. Device authentication is QR-only through the shared WhatsApp setup flow, available from first-run onboarding, the Setup Editor, and estacoda whatsapp. Bridge dependencies stay isolated under scripts/whatsapp-bridge/.

Browser and web debug

Variable	Effect
ESTACODA_BROWSER_DEBUG	Enables browser debug telemetry
ESTACODA_WEB_TOOLS_DEBUG	Enables web tool debug telemetry

Debug data is redacted before storage or return.

Private URL override

Variable	Effect
ESTACODA_ALLOW_PRIVATE_URLS	Overrides security.allowPrivateUrls config. Accepts 1, true, yes, on for true; 0, false, no, off for false. Any other value fails config loading.

Rules

• Do not hardcode secrets in repo files.
• Do not commit real keys.
• Default setup stores pasted secrets in the selected profile .env.
• Advanced setup can reference an existing environment variable by name.
• A custom OpenAI audio env var that is missing does not fall back to OPENAI_API_KEY.
• Resolved key values are never logged or returned in errors.

Related docs

• Configuration — config file families
• State and Files — where .env lives
• Providers — provider setup