Argus Retrieval — Claude Code MCP server: what it does, how to install
Dispatches search across 14 providers, strips web pages down to clean text, and retrieves broken links. Built for agents needing research packs.
claude mcp add argus -e ARGUS_WOLFRAM_API_KEY=<ARGUS_WOLFRAM_API_KEY> -e ARGUS_BRAVE_API_KEY=<ARGUS_BRAVE_API_KEY> -e ARGUS_TAVILY_API_KEY=<ARGUS_TAVILY_API_KEY> -e ARGUS_EXA_API_KEY=<ARGUS_EXA_API_KEY> -e ARGUS_SERPER_API_KEY=<ARGUS_SERPER_API_KEY> -e ARGUS_LINKUP_API_KEY=<ARGUS_LINKUP_API_KEY> -- uvx argus-searchWhat Argus Retrieval does
Argus acts as an access layer between AI agents and web search engines. It connects to 14 search services and skips exhausted budgets automatically. Out of the box, it queries DuckDuckGo and Yahoo without requiring user accounts or tokens. Page extraction follows a 12-stage filtering sequence to deliver clean text instead of raw HTML markup.
Four task modes guide the retrieval process. The tool pulls archived versions of dead links through wayback backends. All fetched data stays inside a designated user directory rather than your local project files.
The setup requires Python 3.11+. Standalone local operation fails without the ARGUS_MCP_STANDALONE=true environment flag. Commercial providers such as Brave or Tavily become active once you pass their respective API keys.
What it helps with
- Pull clean text from articles by URL without boilerplate markup
- Recover deleted or broken articles using web archives
- Build structured documentation packs for target technologies
- Query multiple search providers through a single unified endpoint
Example prompts for the agent
- Search for recent documentation updates on Python web frameworks
- Extract clean article content from
https://example.com/article - Recover the missing webpage at
https://example.com/dead-postusing web archives
How to install Argus Retrieval
Platform CLI
claude mcp add argus -e ARGUS_WOLFRAM_API_KEY=<ARGUS_WOLFRAM_API_KEY> -e ARGUS_BRAVE_API_KEY=<ARGUS_BRAVE_API_KEY> -e ARGUS_TAVILY_API_KEY=<ARGUS_TAVILY_API_KEY> -e ARGUS_EXA_API_KEY=<ARGUS_EXA_API_KEY> -e ARGUS_SERPER_API_KEY=<ARGUS_SERPER_API_KEY> -e ARGUS_LINKUP_API_KEY=<ARGUS_LINKUP_API_KEY> -- uvx argus-searchReplace <…> with your own values. 3 more environment variables are in the server README. uvx runs the command named after the package; if the README names another one, use uvx --from argus-search <command>
Platform CLI
codex mcp add argus-retrieval --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> -- uvx argus-searchReplace <…> with your own values. 3 more environment variables are in the server README. uvx runs the command named after the package; if the README names another one, use uvx --from argus-search <command>
Config file
{
"mcpServers": {
"argus-retrieval": {
"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>"
}
}
}
}Replace <…> with your own values. 3 more environment variables are in the server README. uvx runs the command named after the package; if the README names another one, use uvx --from argus-search <command>
Config file
{
"servers": {
"argus-retrieval": {
"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>"
}
}
}
}Replace <…> with your own values. 3 more environment variables are in the server README. uvx runs the command named after the package; if the README names another one, use uvx --from argus-search <command>
Platform CLI
gemini mcp add -e ARGUS_WOLFRAM_API_KEY=<ARGUS_WOLFRAM_API_KEY> -e ARGUS_BRAVE_API_KEY=<ARGUS_BRAVE_API_KEY> -e ARGUS_TAVILY_API_KEY=<ARGUS_TAVILY_API_KEY> -e ARGUS_EXA_API_KEY=<ARGUS_EXA_API_KEY> -e ARGUS_SERPER_API_KEY=<ARGUS_SERPER_API_KEY> argus-retrieval uvx argus-searchReplace <…> with your own values. 3 more environment variables are in the server README. uvx runs the command named after the package; if the README names another one, use uvx --from argus-search <command>
Step by step, with commands for each platform: How to add an MCP server to Claude Code, Codex, Cursor, VS Code, Gemini CLI and more
Questions about Argus Retrieval
How do I install Argus Retrieval in Claude Code?
Run `claude mcp add argus -e ARGUS_WOLFRAM_API_KEY=<ARGUS_WOLFRAM_API_KEY> -e ARGUS_BRAVE_API_KEY=<ARGUS_BRAVE_API_KEY> -e ARGUS_TAVILY_API_KEY=<ARGUS_TAVILY_API_KEY> -e ARGUS_EXA_API_KEY=<ARGUS_EXA_API_KEY> -e ARGUS_SERPER_API_KEY=<ARGUS_SERPER_API_KEY> -e ARGUS_LINKUP_API_KEY=<ARGUS_LINKUP_API_KEY> -- uvx argus-search`.
What does Argus Retrieval do?
It provides search routing across 14 engines, a 12-stage page text extractor, and historical archive recovery for dead links. It functions immediately with DuckDuckGo and Yahoo before any paid keys are added.
What prerequisites are needed to run Argus locally?
You need Python 3.11 or newer. For standalone development mode, set ARGUS_MCP_STANDALONE=true in your environment.
