DuckDuckGo MCP
Web search through DuckDuckGo with built-in rate limiting, SafeSearch, and content fetching, no API key required.
Delv Safety Grade: B
Score 72/100 · assessed 2026-04-28
DuckDuckGo MCP is a community-maintained search server by solo developer Nick Clyde. It provides read-only web search through DuckDuckGo's API without requiring authentication, which is a genuine privacy advantage. The package is distributed via PyPI and installable through uvx, following standard Python packaging conventions. Permissions are appropriately scoped to outbound network requests for search queries and content fetching. The repository shows reasonable documentation and the code is open source, though it lacks extensive maintenance history or organisational backing. No security incidents are known. The main risk is single-maintainer dependency and the inherent trust placed in DuckDuckGo's API responses, which could theoretically inject malicious content into search results. Rate limiting is built in, reducing abuse potential. Overall, this is a sensible, narrowly-scoped tool with appropriate safety boundaries for its function, but lacks the institutional backing of vendor-maintained alternatives.
Lethal Trifecta (prompt-injection exposure)
TWO OF THREESearch results are arbitrary web content. Outbound to DDG. Add any MCP that touches private state for a complete chain.
Green flags
- No API key required, reducing credential leak surface area
- Read-only search operations with no write permissions
- Built-in rate limiting prevents abuse and API hammering
- Standard PyPI distribution via uvx, not custom install script
- Open source with clear repository and reasonable documentation
Red flags
- Solo maintainer with limited organisational backing or bus factor
- No version pinning visible in install command, potential for supply chain drift
- Search results could inject malicious content from external web sources
Permissions requested
Install
uvx duckduckgo-mcp-server
Review
Install this if you want fast, private web search in Claude without API keys or config hassle. Skip it if you need deep search features, pagination, or Google-level result quality. It's a workhorse for quick fact-checking and research, not a replacement for serious OSINT work.
Good at
- No API key required, so you're running searches in under a minute from install.
- Built-in rate limiting keeps you under DuckDuckGo's radar without manual throttling logic.
- Privacy-first search means no tracking or query history leaking to ad networks.
- Content fetching pulls article text directly, saving you from copying and pasting URLs.
- Works immediately in Claude Desktop with minimal config.
Watch out
- No pagination, so you only see the first page of results and can't dig deeper.
- Content fetcher struggles with JavaScript-heavy sites that rely on client-side rendering.
- DuckDuckGo's result quality is decent but not as comprehensive as Google for niche queries.
- Conservative rate limiting can feel slow if you're doing batch research across multiple topics.
- Manual config needed for hosts beyond Claude Desktop, though it's straightforward.
Use cases
- privacy-first search
- web research
- content summarisation
- fact checking
Getting started
Works with
Similar MCPs
- Brave SearchGive Claude real-time web search via the Brave Search API. The fastest way to add up-to-date knowledge without paying for Perplexity.
- Kagi SearchPrivacy-focused search via Kagi. Great if your team already pays for Kagi and wants Claude to use the same backend.
- DelvSearch the Delv AI tools, MCP, and Skills directory directly from inside your agent. Ask 'what's a good MCP for databases?' in Claude Desktop and get editorial picks, not a random link dump.
- Exa SearchExa neural search for Claude. Particularly good for "find pages similar to X" and research-grade discovery.