Argus is a multi-provider web search broker for AI agents that routes queries across providers like DuckDuckGo, SearXNG, GitHub, Brave, Tavily, and Exa with budget-aware tiering and RRF ranking. It also extracts content from URLs and can recover dead or moved pages via archival and fallback extraction methods.
From the registry: Multi-provider search broker for AI agents: 14 providers, 12-step extraction, retrieval workflows.
$ pipx install argus-search[mcp]Please install the `argus` MCP server into my current AI client (that's you).
Required prerequisites (do these first if not already done):
- **Install Python 3.11+** — Argus requires Python 3.11 or newer. (https://www.python.org/downloads/)
- **Install argus-search with MCP extras** — Install the MCP-enabled Argus package. The README recommends pipx for the MCP use case. Run: `pipx install argus-search[mcp]`
Canonical MCP server config (stdio transport):
- command: `argus`
- args: ["mcp","serve"]
- optional environment variables:
- `ARGUS_SEARXNG_BASE_URL`: SearXNG endpoint for self-hosted search. (example: `<your-searxng-base-url>`)
- `ARGUS_BRAVE_API_KEY`: Brave Search API key. (example: `<your-brave-api-key>`)
- `ARGUS_SERPER_API_KEY`: Serper API key. (example: `<your-serper-api-key>`)
- `ARGUS_TAVILY_API_KEY`: Tavily API key. (example: `<your-tavily-api-key>`)
- `ARGUS_EXA_API_KEY`: Exa API key. (example: `<your-exa-api-key>`)
- `ARGUS_LINKUP_API_KEY`: Linkup API key. (example: `<your-linkup-api-key>`)
- `ARGUS_PARALLEL_API_KEY`: Parallel AI API key. (example: `<your-parallel-api-key>`)
- `ARGUS_YOU_API_KEY`: You.com API key. (example: `<your-you-api-key>`)
- `ARGUS_VALYU_API_KEY`: Valyu API key for search, contents, and answer features. (example: `<your-valyu-api-key>`)
- `ARGUS_FIRECRAWL_API_KEY`: Firecrawl API key for content extraction. (example: `<your-firecrawl-api-key>`)
- `ARGUS_GITHUB_API_KEY`: GitHub token for higher rate limits. (example: `<your-github-token>`)
- `ARGUS_*_MONTHLY_BUDGET_USD`: Optional per-provider monthly budget limit in USD. (example: `<your-monthly-budget-usd>`)
- `ARGUS_CRAWL4AI_ENABLED`: Enable Crawl4AI extraction step. (example: `<true-or-false>`)
- `ARGUS_YOU_CONTENTS_ENABLED`: Enable You.com Contents API extraction. (example: `<true-or-false>`)
- `ARGUS_CACHE_TTL_HOURS`: Result cache TTL in hours. (example: `<cache-ttl-hours>`)
Note: Default MCP setup is stdio with zero required environment variables; Argus works out of the box using DuckDuckGo. The README also documents an SSE mode for self-hosted remote access via `argus mcp serve --transport sse --host 0.0.0.0 --port 8001`, but the documented client URL example points to `http://<your-server>:8271/sse`, which is inconsistent with the shown server port.
Add this MCP server to my current client's config in the correct format for you. If you need secrets or credentials I haven't provided, ASK me — do not invent values or leave raw placeholders. After adding it, tell me how to verify the server is connected.MCP server for searching Airweave collections with natural language queries.