Web Retrieval Mcp
Give your agent SSRF-guarded Exa search and Firecrawl web fetch via stdio MCP on Windows, macOS, or Linux.
Overview
web-retrieval-mcp is a MCP server for the Idea phase that combines Exa search and Firecrawl tiered fetch with SSRF guards for agent web research.
What is this MCP server?
- Combines Exa web search with tiered web fetch (Firecrawl) in one PyPI package web-retrieval-mcp 0.1.2
- Advertises SSRF guarding for safer agent-driven URL retrieval
- Cross-platform stdio transport—fits local Claude Code and Cursor setups
- Requires EXA_API_KEY and FIRECRAWL_API_KEY (both document free tiers at exa.ai and firecrawl.dev)
- Published on PyPI with identifier web-retrieval-mcp
- Package version 0.1.2 on PyPI
- 2 required secret environment variables: EXA_API_KEY, FIRECRAWL_API_KEY
- stdio transport; repository github.com/VelvetSP/web-retrieval-mcp
What problem does it solve?
Agents default to stale training data or unsafe fetches when builders need current pages and search results inside the IDE.
Who is it for?
Indie developers who already use Exa and Firecrawl and want one MCP package for search-plus-read workflows during discovery and validation.
Skip if: Builders who cannot use external APIs, need logged-in browser sessions only, or want a zero-key local crawler.
What do I get? / Deliverables
After installing the PyPI server and API keys, your agent can search the web and retrieve page content through guarded, tiered fetch tools.
- Working stdio MCP server exposing search and tiered fetch tools
- Retrieved markdown or text from target URLs without ad-hoc scraping scripts
- Repeatable research loop the agent can call from idea through launch tasks
Recommended MCP Servers
Journey fit
Tiered web search and fetch is primarily a research primitive—agents need grounded pages before positioning, SEO, or competitive calls. Research is the canonical shelf for Exa-powered search plus Firecrawl-backed page retrieval in one MCP.
How it compares
SSRF-aware search-and-fetch MCP integration, not a passive skills.sh documentation skill.
Common Questions / FAQ
Who is Web Retrieval Mcp for?
Web Retrieval Mcp is for builders running MCP clients who need Exa search and Firecrawl fetches with SSRF guarding during product research and competitive analysis.
When should I use Web Retrieval Mcp?
Use it whenever your agent must query the live web and pull page text safely—especially in idea research, competitor validation, and launch content work—not for editing your local repo.
How do I add Web Retrieval Mcp to my agent?
Install Web Retrieval Mcp from PyPI, set EXA_API_KEY and FIRECRAWL_API_KEY, then add a stdio MCP server entry with transport type stdio per your client’s config docs.