MCP Guide
Validate hotel feeds with an AI agent
Connect the metasearch remote MCP endpoint and let an agent detect formats, normalize data, validate platforms, compare requirements and suggest fixes through tools.
MCPStreamable HTTP
Endpoint
https://metasearch.com.tr/mcpClaude Code
claude mcp add --transport http metasearch https://metasearch.com.tr/mcp
claude mcp listUse Claude Code's project MCP scope when the configuration should be shared with a repository.
Cursor
.cursor/mcp.json
{
"mcpServers": {
"metasearch": {
"url": "https://metasearch.com.tr/mcp"
}
}
}VS Code / GitHub Copilot
.vscode/mcp.json
{
"servers": {
"metasearch": {
"type": "http",
"url": "https://metasearch.com.tr/mcp"
}
}
}Recommended agent workflow
1. Detect + normalize
detect_feed_format → normalize_hotel_feed
2. Compare + validate
compare_platform_requirements → validate_feed
3. Fix
suggest_feed_fixes → fix source feed → validate again
Prompt recipes
Validate supplier_hotels.csv for Google Hotel Center. Group errors by issue code, show affected rows, and suggest the minimum fixes.Compare this feed for Google Hotel Center, Wego and trivago. Separate shared canonical fields from platform-specific gaps.Detect the feed format first, normalize it to the canonical hotel model, then produce a readiness score and fix list for each target.When to use MCP vs CLI
MCP
For interactive agent analysis, explanations, platform comparison and remediation.
@metasearch/feed-validator
For local files, public URLs, stdin and CI/CD validation.
Data safety
Do not place credentials, API keys, payment data or other sensitive information in a feed. MCP validation payloads are not intentionally persisted, but send only the hotel/catalog data required for validation.