type: rule
status: active
timestamp: 2026-06-27
tags: [mcp, search, fallback, no-card, no-key]
status: active
timestamp: 2026-06-27
tags: [mcp, search, fallback, no-card, no-key]
Web search — 3-MCP fallback chain (10 engines, no keys)
searxng + duckduckgo + open-websearch. 10 engines, no API key. Try in order on failure
Web search — 3-MCP fallback chain
Rule
Three no-key web-search MCPs always installed at user scope. Try them in priority order:
| Priority | MCP | Engines | Notes |
|---|---|---|---|
| 1 | searxng | Google + Bing + Wikipedia (meta) | Best aggregate quality |
| 2 | open-websearch | Bing, Baidu, DuckDuckGo, Brave, Exa, GitHub, Juejin, CSDN | 8 engines, AUTO routing |
| 3 | duckduckgo | DuckDuckGo direct | Simplest fallback |
Total engine coverage: 10 distinct engines across the 3 MCPs. All free. All zero-API-key. No card-on-file.
Skipped (require API key)
- Brave Search MCP (free 2k/mo) — duplicate, no-key open-websearch covers Brave
- Tavily (free 1k/mo) — no-key alternatives exist
- Exa ($10 free credit) — open-websearch covers Exa
- Kagi (paid + card) — violates no-card-on-file
Revisit if all 3 no-key MCPs fail simultaneously on a query.
How to apply
When you need web search:
- Try searxng tool first
- On failure (rate-limit, 403, 5xx, empty results) ? open-websearch
- On second failure ? duckduckgo
- Only escalate (ask user / add new MCP) if all 3 fail consistently
Per try-multiple-on-failure — at least 3 alts before reporting blocker. This rule MAKES that achievable in one click.