google-cli

google-cli

CLI wrapper around Google Search backed by the kimi-webbridge browser daemon. Runs inside the user's real Chrome session — no API key, no login loop — and emits JSON on stdout.

Commands

Command	Usage	Returns
search	google-cli search <query> [--limit 10] [--hl en]	[{title, url, snippet}]
result	google-cli result <url>	{url, title, description, text}

All output:

{"ok": true, "data": ...}

or on failure (non-zero exit):

{"ok": false, "error": {"code": "...", "message": "..."}}

Prerequisites

1. kimi-webbridge daemon running on 127.0.0.1:10086 — install from https://www.kimi.com/features/webbridge.
2. Go 1.25+ for building.

Build

go build -o google-cli .
./google-cli --help

Quick test

./google-cli search "claude code" --limit 3
./google-cli result "https://code.claude.com/docs/en/overview"

Consent pages

EU / anonymous browsers may hit consent.google.com on first Google navigation. When detected, search errors with code consent_required — accept it once in Chrome, then retry.

How selectors stay alive

Google frequently renames DOM classes. The working selectors, rationale, and failure modes are pinned in ARCHAEOLOGY.md. When parsing breaks, re-run the site-archaeology protocol from that file and update both google/search.go and ARCHAEOLOGY.md together.

Layout

google-cli/
├── main.go                # entrypoint
├── browser/client.go      # kimi-webbridge HTTP client (+ EvaluateJSON helper)
├── output/output.go       # JSON contract: {ok, data} / {ok, error}
├── google/
│   ├── common.go          # evaluateWithRetry (handles CDP context races)
│   ├── search.go          # search command backend
│   └── result.go          # result command backend
├── cmd/
│   ├── root.go
│   ├── search.go
│   └── result.go
├── ARCHAEOLOGY.md         # DOM-selector field notes
└── README.md

License

MIT (see LICENSE — to be added).