Skip to content
MCP server

Argus Retrieval

Search broker over 14 providers with page extraction, dead-link recovery, research packs

Runs locallyStatus and history
Install
Pricing
Open source
Platforms
CLI, API, self-hosted
Runs
Locally
Version
1.6.5
Listing updated
today

Similar MCP servers

About Argus Retrieval

Argus Retrieval is an open-source MCP server for knowledge and memory. It runs locally on your machine and is started by your AI client.

Argus began as a way to route one search query across many providers, free ones first, skipping any whose budget is spent. It starts with no accounts at all, using DuckDuckGo and Yahoo, and grows by adding keys for services like Brave, Tavily, Exa, Serper, Linkup or WolframAlpha, or a self-hosted SearXNG that aggregates over 70 engines. Results are fused with reciprocal rank fusion, and scores can show which providers contributed.

Beyond search it extracts full page text through a 12-step pipeline with quality gates, and offers workflows that write local artifacts: recover a dead article using the Wayback Machine and archives, capture the important pages of a site, and build a docs-plus-research pack with citations. It has four modes, discovery, research, recovery and grounding, plus sessions for multi-turn context.

The README is candid that it is beta and posts a dated status note: production search works through SearXNG, Yahoo and GitHub, DuckDuckGo is intermittent, and paid providers are not currently admitted. It can run as an HTTP API, CLI, MCP server or Python SDK, with a usage dashboard.

Install

Run in a terminal

claude mcp add argus --env ARGUS_WOLFRAM_API_KEY='<ARGUS_WOLFRAM_API_KEY>' --env ARGUS_BRAVE_API_KEY='<ARGUS_BRAVE_API_KEY>' --env ARGUS_TAVILY_API_KEY='<ARGUS_TAVILY_API_KEY>' --env ARGUS_EXA_API_KEY='<ARGUS_EXA_API_KEY>' --env ARGUS_SERPER_API_KEY='<ARGUS_SERPER_API_KEY>' --env ARGUS_LINKUP_API_KEY='<ARGUS_LINKUP_API_KEY>' -- uvx argus-search

Add to claude_desktop_config.json

{
  "mcpServers": {
    "argus": {
      "command": "uvx",
      "args": [
        "argus-search"
      ],
      "env": {
        "ARGUS_WOLFRAM_API_KEY": "<ARGUS_WOLFRAM_API_KEY>",
        "ARGUS_BRAVE_API_KEY": "<ARGUS_BRAVE_API_KEY>",
        "ARGUS_TAVILY_API_KEY": "<ARGUS_TAVILY_API_KEY>",
        "ARGUS_EXA_API_KEY": "<ARGUS_EXA_API_KEY>",
        "ARGUS_SERPER_API_KEY": "<ARGUS_SERPER_API_KEY>",
        "ARGUS_LINKUP_API_KEY": "<ARGUS_LINKUP_API_KEY>"
      }
    }
  }
}

Add to ~/.cursor/mcp.json

{
  "mcpServers": {
    "argus": {
      "command": "uvx",
      "args": [
        "argus-search"
      ],
      "env": {
        "ARGUS_WOLFRAM_API_KEY": "<ARGUS_WOLFRAM_API_KEY>",
        "ARGUS_BRAVE_API_KEY": "<ARGUS_BRAVE_API_KEY>",
        "ARGUS_TAVILY_API_KEY": "<ARGUS_TAVILY_API_KEY>",
        "ARGUS_EXA_API_KEY": "<ARGUS_EXA_API_KEY>",
        "ARGUS_SERPER_API_KEY": "<ARGUS_SERPER_API_KEY>",
        "ARGUS_LINKUP_API_KEY": "<ARGUS_LINKUP_API_KEY>"
      }
    }
  }
}

Add to .vscode/mcp.json

{
  "servers": {
    "argus": {
      "type": "stdio",
      "command": "uvx",
      "args": [
        "argus-search"
      ],
      "env": {
        "ARGUS_WOLFRAM_API_KEY": "<ARGUS_WOLFRAM_API_KEY>",
        "ARGUS_BRAVE_API_KEY": "<ARGUS_BRAVE_API_KEY>",
        "ARGUS_TAVILY_API_KEY": "<ARGUS_TAVILY_API_KEY>",
        "ARGUS_EXA_API_KEY": "<ARGUS_EXA_API_KEY>",
        "ARGUS_SERPER_API_KEY": "<ARGUS_SERPER_API_KEY>",
        "ARGUS_LINKUP_API_KEY": "<ARGUS_LINKUP_API_KEY>"
      }
    }
  }
}

Add to ~/.codeium/windsurf/mcp_config.json

{
  "mcpServers": {
    "argus": {
      "command": "uvx",
      "args": [
        "argus-search"
      ],
      "env": {
        "ARGUS_WOLFRAM_API_KEY": "<ARGUS_WOLFRAM_API_KEY>",
        "ARGUS_BRAVE_API_KEY": "<ARGUS_BRAVE_API_KEY>",
        "ARGUS_TAVILY_API_KEY": "<ARGUS_TAVILY_API_KEY>",
        "ARGUS_EXA_API_KEY": "<ARGUS_EXA_API_KEY>",
        "ARGUS_SERPER_API_KEY": "<ARGUS_SERPER_API_KEY>",
        "ARGUS_LINKUP_API_KEY": "<ARGUS_LINKUP_API_KEY>"
      }
    }
  }
}

Add to your client's MCP settings (most use this shape)

{
  "mcpServers": {
    "argus": {
      "command": "uvx",
      "args": [
        "argus-search"
      ],
      "env": {
        "ARGUS_WOLFRAM_API_KEY": "<ARGUS_WOLFRAM_API_KEY>",
        "ARGUS_BRAVE_API_KEY": "<ARGUS_BRAVE_API_KEY>",
        "ARGUS_TAVILY_API_KEY": "<ARGUS_TAVILY_API_KEY>",
        "ARGUS_EXA_API_KEY": "<ARGUS_EXA_API_KEY>",
        "ARGUS_SERPER_API_KEY": "<ARGUS_SERPER_API_KEY>",
        "ARGUS_LINKUP_API_KEY": "<ARGUS_LINKUP_API_KEY>"
      }
    }
  }
}

Using several servers? Build one config file with the MCP config generator.

Configuration

  • ARGUS_WOLFRAM_API_KEYWolframAlpha App ID (2,000 free computed-answer queries/month)
  • ARGUS_BRAVE_API_KEYBrave Search API key (2,000 free queries/month)
  • ARGUS_TAVILY_API_KEYTavily API key (1,000 free queries/month)
  • ARGUS_EXA_API_KEYExa API key (1,000 free queries/month)
  • ARGUS_SERPER_API_KEYSerper API key (2,500 one-time free queries)
  • ARGUS_LINKUP_API_KEYLinkup API key (1,000 free queries/month)
  • ARGUS_SEARXNG_BASE_URLSearXNG instance URL (self-hosted, aggregates 70+ engines)
  • ARGUS_DATA_ROOTOptional override for the writable Argus runtime corpus root

What it does

14 search providers
Routes queries free-first and skips providers that are out of budget.
Content extraction
Returns full page text through a 12-step pipeline with quality checks.
Dead URL recovery
Finds archived copies through Wayback Machine and fallbacks.
Site capture
Saves the important on-domain pages with a cited summary.
Research packs
Builds local docs-plus-research packs from official URLs.
Zero-key start
Works with DuckDuckGo and Yahoo before any API key is added.

Good for

  • Give an agent web search without paying for one API
  • Recover an article that has disappeared
  • Capture a documentation site for offline reference
  • Build a research pack on an SDK
  • Compare how several search engines rank a query

Pricing

Open source

Open source; optional provider keys have their own free quotas

Argus Retrieval FAQ

Do I need API keys?

No. A plain pip install gives DuckDuckGo and Yahoo; keys for other providers are optional.

Where does it store data?

In a writable user data directory, or ARGUS_DATA_ROOT, not in the repository checkout.

Is it production ready?

The README labels it beta and publishes a dated status matrix to check before relying on a provider.

How can I use it?

Through an HTTP API, CLI, MCP server or Python SDK.

Is Argus Retrieval free?

Yes. Argus Retrieval is open source, so you can use and self-host it for free.

How do I add Argus Retrieval to Claude Code?

Run this command in your terminal: claude mcp add argus --env ARGUS_WOLFRAM_API_KEY='<ARGUS_WOLFRAM_API_KEY>' --env ARGUS_BRAVE_API_KEY='<ARGUS_BRAVE_API_KEY>' --env ARGUS_TAVILY_API_KEY='<ARGUS_TAVILY_API_KEY>' --env ARGUS_EXA_API_KEY='<ARGUS_EXA_API_KEY>' --env ARGUS_SERPER_API_KEY='<ARGUS_SERPER_API_KEY>' --env ARGUS_LINKUP_API_KEY='<ARGUS_LINKUP_API_KEY>' -- uvx argus-search

What are the best alternatives to Argus Retrieval?

The closest alternatives to Argus Retrieval are Scrapling MCP Server, Codebase Memory, XERJ, Firecrawl MCP Server and Claude Code Ultimate Guide.

Where this comes from

  • Description: written from the project's README.
  • Install steps: from the official MCP Registry entry.

How AIPoweredBot checks servers