Search the web using Google Search via Gemini grounding — current events, facts, documentation
Check: which gski
Install if missing: pip install -e . from ~/Documents/gski
Requires GEMINI_API_KEY env var.
gski websearch sends a query to Gemini with Google Search grounding enabled. The model searches the web, synthesizes results, and returns text with inline citation references and resolved source URLs.
Vertex AI Search redirect URLs are automatically resolved to actual source URLs.
Default model is gemini-2.5-flash.
# Basic search
gski websearch "who won the 2024 world series"
# Current events
gski websearch "latest news on AI regulation"
# Technical lookup
gski websearch "python 3.13 new features"
# Use flash-lite for cheaper/faster queries
gski websearch "weather in berlin" --model flash-lite
# Raw output (no citation formatting)
gski websearch "euro 2024 results" --raw
| Flag | Values | Default | Notes |
|---|---|---|---|
--model | flash, flash-lite | flash | model selection |
--raw | flag | off | plain text without citation formatting |
Default output includes:
[1], [2]Sources: section with numbered URLs and titlesWith --raw: just the plain response text, no citations or sources.
Use when you need information beyond your training data: current events, recent releases, live data, verifying claims, documentation lookups.