google-ai-mode-mcp
Provides an MCP server tool (search_ai) that automates using Google AI Mode in a local browser session, extracts the synthesized answer and citations, and returns them as Markdown plus source links. Includes CAPTCHA handling, language handling, and persistence of a browser profile.
Score Breakdown
⚙ Agent Friendliness
🔒 Security
Runs locally with a persistent browser profile; no explicit secret handling is described (no API keys). However, it employs stealth techniques and browser automation against Google, which can increase security/abuse risk and may require careful isolation of the runtime environment. CAPTCHA handling may require headful mode, increasing operator exposure. Dependency hygiene and supply-chain posture are not verifiable from the provided manifest/README alone.
⚡ Reliability
Best When
You want token-efficient, cited research inside an MCP-compatible agent and can tolerate occasional CAPTCHA/retry behavior.
Avoid When
You need guaranteed uptime, strong security assurances, or you cannot run/inspect an automated browser profile.
Use Cases
- • Web research with inline citations for technical topics (frameworks, databases, coding patterns)
- • Comparative research (e.g., PostgreSQL vs MySQL) with sourced claims
- • Regulatory/news research where citations are required
- • Product/spec comparison summaries with linked sources
- • General “ask and get cited answer” workflows in MCP-compatible code agents
Not For
- • Scenarios requiring strict compliance with a provider’s documented API terms (it relies on browser automation/stealth)
- • Use cases needing high availability or deterministic results
- • Environments where running a stealth browser is disallowed
- • Cases that require a formal REST/SDK integration contract beyond MCP tool calls
Interface
Authentication
No API keys/auth flows are described. Authentication is effectively delegated to the user’s local Google/browser session and any CAPTCHA interaction.
Pricing
Cost is primarily local compute/browser automation plus your agent runtime. Network cost is via Google access; no paid plan details are provided.
Agent Metadata
Known Gotchas
- ⚠ CAPTCHA may occur on first run or frequent searches; agent may need to run with headless:false and wait for manual CAPTCHA completion.
- ⚠ The server forces/assumes certain language behavior; if language is wrong, it suggests clearing the browser profile.
- ⚠ Browser profile state is important; clearing profile may be required after failures or repeated CAPTCHAs.
- ⚠ Browser display/X-server issues on Linux/WSL may require xvfb-run to make CAPTCHA solving possible.
- ⚠ Since results are produced by browser automation, transient failures may occur (stealth detection, layout changes), requiring retries via updating the package and/or restarting the browser context.
Alternatives
Full Evaluation Report
Comprehensive deep-dive: security analysis, reliability audit, agent experience review, cost modeling, competitive positioning, and improvement roadmap for google-ai-mode-mcp.
AI-powered analysis · PDF + markdown · Delivered within 30 minutes
Package Brief
Quick verdict, integration guide, cost projections, gotchas with workarounds, and alternatives comparison.
Delivered within 10 minutes
Score Monitoring
Get alerted when this package's AF, security, or reliability scores change significantly. Stay ahead of regressions.
Continuous monitoring
Scores are editorial opinions as of 2026-03-30.