Searxng Deepdive
Let your agent run privacy-friendly SearXNG searches with engine targeting, multi-page fanout, and full-page Markdown reads for deep research without manual tab hopping.
Overview
SearXNG Deepdive is a MCP server for the Idea phase that powers agent-driven SearXNG search with engine targeting, multi-page fanout, and HTML-to-Markdown page reading.
What is this MCP server?
- stdio npm package searxng-deepdive v0.3.3 for local MCP wiring
- Engine targeting and multi-page fanout beyond a single SERP snippet
- HTML→Markdown URL reading for agent-friendly long-context ingestion
- SEARXNG_URL env var points at your instance (default http://127.0.0.1:8080 if unset; docs suggest e.g. http://127.0.0.1:
- Self-hosted SearXNG required—you control engines and privacy posture
- Package version 0.3.3
- 1 required environment variable: SEARXNG_URL
- Transport: stdio via npm identifier searxng-deepdive
What problem does it solve?
Agent research stalls on shallow snippets and copy-paste from browsers, with no repeatable way to deep-read sources into the session.
Who is it for?
Privacy-conscious solo builders who already run or will run SearXNG and want MCP-native deep research in the agent loop.
Skip if: Builders who want managed web search with zero self-hosting or no appetite to maintain a SearXNG instance.
What do I get? / Deliverables
After npm stdio setup and SEARXNG_URL configuration, your agent can target engines, fan out pages, and ingest full articles as Markdown for analysis.
- Targeted multi-engine search results inside agent sessions
- Paginated fanout across result sets
- Markdown renditions of fetched HTML pages for summarization and citation
Recommended MCP Servers
Journey fit
Idea phase is the canonical shelf because the server's core loop is discovery and synthesis—competitors, markets, and technical unknowns—before you commit to build. Research subphase fits engine-targeted search plus HTML→Markdown URL reading for evidence gathering agents rely on during validation.
How it compares
Self-hosted meta-search MCP with page extraction, not a proprietary hosted SERP API skill.
Common Questions / FAQ
Who is searxng-deepdive for?
Indie builders and agent users who self-host SearXNG and need structured, multi-page search plus Markdown page reads inside MCP clients.
When should I use searxng-deepdive?
Use it during competitor research, market validation, and technical discovery when you need more than one page of results and full article text in the agent context.
How do I add searxng-deepdive to my agent?
Install the npm package searxng-deepdive (stdio), set SEARXNG_URL to your running SearXNG base URL, and register the server in your MCP client configuration.