# obsidian-second-brain — research toolkit credentials
# Copy this file to ~/.config/obsidian-second-brain/.env and fill in the values.
# That path is Mac-local — never committed, never synced. Set permissions 600.

# ─── xAI Grok ───────────────────────────────────────────────
# Powers /x-read and /x-pulse (live X reading + trend analysis)
# Get key: https://console.x.ai
XAI_API_KEY=

# ─── Perplexity ─────────────────────────────────────────────
# Powers /research and /research-deep (web research with citations)
# Get key: https://perplexity.ai/settings/api
PERPLEXITY_API_KEY=

# ─── YouTube Data API v3 (optional) ─────────────────────────
# Powers /youtube metadata + comments. Transcripts work without this key.
# Get key: https://console.cloud.google.com → enable YouTube Data API v3
YOUTUBE_API_KEY=

# ─── Model defaults (override if desired) ───────────────────
# xAI endpoint: https://api.x.ai/v1/responses (Agent Tools API)
# grok-4 (non-reasoning) is the right default for /x-read /x-pulse /youtube — these are
# extract-and-format tasks where reasoning is wasted overhead (3-4x more output tokens).
# Switch to grok-4.20-reasoning per-call if you want deeper analysis.
GROK_MODEL=grok-4
PERPLEXITY_RESEARCH_MODEL=sonar-pro
PERPLEXITY_DEEP_MODEL=sonar-deep-research

# ─── Auto-open behavior ─────────────────────────────────────
# After every research save, Obsidian opens automatically at the note.
# Set to 0 to disable (e.g. when running batch saves or scheduled).
# RESEARCH_AUTOOPEN=0