.env.example

# Conference Paper Analyzer — environment variables
#
# Copy this file to `.env` and fill in the key(s) for the provider you use:
#
#     cp .env.example .env
#
# The app loads `.env` automatically on startup. Every value is optional — set
# only what your chosen LLM provider needs. The in-app "API key" / "LLM endpoint"
# fields override these at runtime.

# ── Anthropic / Claude (default provider) ───────────────────────────────────
ANTHROPIC_API_KEY=
# Optional: custom Anthropic-compatible base URL
# ANTHROPIC_BASE_URL=

# ── OpenAI (install:  uv sync --extra openai) ───────────────────────────────
OPENAI_API_KEY=
# Optional: OpenAI-compatible base URL (the OpenAI SDK reads this automatically;
# you can also set it per-run in the UI's "LLM endpoint" field)
# OPENAI_BASE_URL=

# ── LiteLLM (install:  uv sync --extra litellm) ─────────────────────────────
# Key for your LiteLLM proxy/endpoint. If unset, LITELLM falls back to OPENAI_API_KEY.
# Set the endpoint itself in the UI's "LLM endpoint" field.
# LITELLM_API_KEY=
#
# LiteLLM also reads provider-native keys from the environment when it routes to
# a given backend, e.g.:
# AZURE_API_KEY=
# AZURE_API_BASE=
# GEMINI_API_KEY=
# COHERE_API_KEY=

# ── OpenReview source (ICLR / NeurIPS) ──────────────────────────────────────
# Usually required. OpenReview now gates anonymous note queries behind an
# anti-bot challenge (HTTP 403), so ICLR / NeurIPS venues need a logged-in
# session. Provide credentials to authenticate (this also enables private
# access). Use EITHER a pre-obtained token OR username + password.
# OPENREVIEW_TOKEN=
# OPENREVIEW_USERNAME=
# OPENREVIEW_PASSWORD=