Argus Retrieval
Centralize multi-provider web and knowledge search behind one MCP broker so your agent can run structured 12-step retrieval instead of guessing which API to call.
Overview
Argus is a MCP server for the Idea phase that brokers 14 search providers and a 12-step extraction workflow so AI agents can run structured retrieval research.
What is this MCP server?
- Multi-provider search broker with 14 providers in catalog positioning
- Documented 12-step extraction workflow for retrieval-oriented agent tasks
- PyPI package argus-search with uvx runtime hint and stdio MCP transport
- Optional API keys for Wolfram, Brave, Tavily, Exa, Serper, Linkup with documented free-tier query allowances
- Argus Retrieval MCP version 1.6.2 from io.github.Khamel83/argus
- 14 search providers referenced in catalog description
- 12-step extraction workflow referenced in catalog description
- MCP server version 1.6.2
Community signal: 2 GitHub stars.
What problem does it solve?
Research agents waste time picking search APIs, juggling keys, and improvising extraction steps instead of following one retrieval workflow.
Who is it for?
Indie builders doing competitive scans, technical lookups, and fact gathering across several search APIs from Claude Code or Cursor.
Skip if: Teams that need a single locked-in search vendor with no key management, or purely internal document RAG with no web providers.
What do I get? / Deliverables
After you connect Argus, your agent can query multiple search backends through one MCP surface with a consistent multi-step extraction path.
- Unified MCP search and extraction across configured third-party providers
- Agent-ready retrieval workflows aligned to the documented 12-step extraction model
- Configurable provider mix without rewriting agent prompts per vendor
Recommended MCP Servers
Journey fit
Argus is a research and retrieval broker you reach for when exploring markets, APIs, and facts before you commit to a build—canonical Idea → research. It orchestrates searches and extraction across many providers, which is competitive and technical research—not shipping, monitoring, or billing.
How it compares
Multi-provider retrieval MCP broker, not a standalone SEO skill or a hosted analytics product.
Common Questions / FAQ
Who is Argus for?
It is for developers who want one MCP layer that fans out to many search and knowledge providers during agent-led research.
When should I use Argus?
Use it in early idea and validation research when you need citations, comparisons, or computed answers before you design features or pick integrations.
How do I add Argus to my agent?
Install/run argus-search from PyPI with uvx, configure stdio MCP in your agent host, and set ARGUS_* API key environment variables for the providers you enable.