OpenSERP

Multi-engine search One search endpoint queries Google, Bing, Yandex, Yahoo, Baidu, DuckDuckGo, SearXNG, Brave, Kagi, Mojeek, Ecosia, Marginalia, Exa, YouTube, eBay, crates.io, and npmjs, with per-engine pages, time-range, and safe-search parameters.[1][2][3]

Structured JSON results Searches return typed JSON with titles, URLs, snippets, positions, and engine-specific fields such as knowledge-graph panels, related searches, and People Also Ask.[1][2]

Full-page scraping A dedicated scrape endpoint fetches any URL and returns page content as markdown, keeping a search-then-read workflow inside one API and key.[1][2]

AI-ready endpoints Natural-language AI search takes a query plus a JSON schema and returns schema-conformant answers via SearXNG, Exa, OpenAI, or Gemini providers.[1][2]

MCP server for agents A hosted MCP endpoint with install commands for Claude Code, Cursor, and VS Code lets coding agents call web search without custom glue code.[1][2]

Keys, quotas, and dashboard The web dashboard issues API keys, assigns monthly quotas per key, and reports usage, logs, latency, and per-provider analytics.[1][2]

One-command self-hosting The stack deploys with one Docker Compose command; the README documents the API, SQLite, Redis, Playwright, and Chromium services plus the SearXNG setup step.[1][2]

Free core with usage-based cloud Self-hosting is free forever under the MIT license; the hosted cloud bills 1 credit per search, 2 per scrape, and 5 per AI search at $0.0078 per credit with 60 free credits to start.[1][2][3]

Agent-building indie developers Indie developers wiring search into AI agent loops fit well because the MCP server, schema-bound AI search, and SDKs remove glue code.[1][2]

Budget SEO tool builders SEO tool builders on a budget fit well because multi-engine results with positions, related searches, and People Also Ask cover rank-tracking inputs without per-seat SaaS pricing.[1][2]

Self-hosting privacy-minded teams Privacy-minded teams that want queries on infrastructure they control fit well because MIT-licensed self-hosting keeps data in local SQLite and Redis.[1][2]

No uptime or support guarantees Teams needing guaranteed uptime and support should avoid it: no SLA, enterprise tier, or support plan is documented.[1][2]

Developer-only operation Non-technical users wanting a no-code product should avoid it: deployment needs Docker plus a SearXNG instance and consumption needs API integration.[1][2]

Undocumented compliance posture Buyers needing compliance paperwork should avoid it: data residency, SSO/SAML, and security attestations are undocumented.[1][2]