.env.example

# Template for real secrets used when running docs/999-usage-scenarios.md scenarios against a real
# backend instead of MockBackend. Copy to .env (already gitignored -- never commit real values)
# and `set -a && source .env && set +a` before running a scenario.
#
# Candle needs nothing here at all -- it's a local model with no auth.

# Local engines (ollama/vllm/litellm/custom) -- read directly by hyperion-console at
# `/backend <name> <model>` time. All optional: Ollama and vLLM usually don't need one; a
# self-hosted LiteLLM proxy often does.
HYPERION_OLLAMA_API_KEY=
HYPERION_VLLM_API_KEY=
HYPERION_LITELLM_API_KEY=
HYPERION_CUSTOM_API_KEY=

# Cloud providers are NOT read from the environment by hyperion-console -- there is no
# HYPERION_OPENAI_API_KEY equivalent. The console only accepts a cloud key through its own real
# "connect my <provider> account" utterance flow, which stores it encrypted at rest. These
# variables exist only so a *scenario script* can pipe them in as if you'd typed them -- see
# docs/999-usage-scenarios.md's "Running a scenario against a real backend" section.
OPENAI_API_KEY=
ANTHROPIC_API_KEY=
GEMINI_API_KEY=
GROQ_API_KEY=