The world's search gateway for AI agents & developers
One endpoint in front of a weighted pool of search engines. Your agents get a single normalized response, one predictable bill, and zero fan-out.
POST /api/search · bearer authPOST /api/search returns — move your key to go live.
{ "provider": "…", "results": [ … ] }
Built for production agent traffic
Four endpoints. Every retrieval pattern.
Search, read, combine, or parse. Each call is routed to one provider and normalized to a single schema, so switching providers never breaks your app.
Everything a search gateway should do
Built for agents that need fresh web data without stitching together four vendor SDKs and four billing dashboards.
Weighted round-robin gateway
Automatic load balancing across every configured provider, with free providers carrying a larger share — one call, one bill, no fan-out.
Instant extraction
Fast page reading and content parsing through /api/read and /api/extract.
Real-time telemetry
Track every query, status code and latency. Postgres-backed, visible in the dashboard.
Secure bearer auth
Production-ready API keys, hashed at rest, revocable per project in one click.
Where Meldrix fits
Anywhere an agent needs the live web, without vendor lock-in or unpredictable spend.
Autonomous research agents
Give your agent one search tool with consistent results and a predictable cost per step, even when a provider degrades.
- /api/search
- /api/read
Retrieval pipelines
Search and scrape in one call, then chunk the normalized content straight into your vector store.
- /api/search-and-read
- /api/extract
Copilots & dashboards
Ship a search feature without embedding four SDKs or reconciling four invoices every month.
- Bearer keys
- Telemetry
One request, one provider. Pay for what you route.
Usage-based pricing across Parallel, Exa, Geekflare, Serper, Tavily and LangSearch. Create a key, watch the live telemetry, and scale when your agents do.