web3-research

Web3 Research MCP is a local, free-form research assistant focused on gathering and organizing information about cryptocurrency tokens. It provides a set of tools to plan, search, and document research across multiple sources (e.g., CoinGecko, CoinMarketCap, DeFiLlama, and more). Core capabilities include creating structured research plans for tokens, performing web searches, conducting keyword-targeted queries, updating the progress of different research sections, fetching and saving content as resources, and generating comprehensive reports. This MCP server is intended to run locally via npx and can be integrated with Claude Desktop or Cursor to streamline token research workflows.

Prerequisites:

• Node.js (v16 or higher) and npm/yarn installed on your system
• Internet access to install the MCP package

Installation steps:

1. Install the MCP via Smithery (recommended):

   npx -y @smithery/cli install web3-research-mcp --client claude
   

2. Start using the MCP in Claude Desktop or Cursor by configuring the MCP server as shown below.
3. If you prefer manual installation, you can install the MCP package globally in your project context via npm and run it with npx as needed:

   npm install -g web3-research-mcp@latest
   # or use npx directly
   npx -y web3-research-mcp@latest
   

4. Verify installation by starting the MCP and ensuring it responds to basic commands through your integration (Claude Desktop or Cursor).

Note: Ensure your environment allows network access for multi-source searches and content fetching; some sources may block scraping or return 403 errors occasionally.

Tips:

• The MCP relies on online data sources; some sites may block automated access or return partial results. If you encounter 403s, try alternate sources or adjust the search parameters.
• When using with Claude Desktop or Cursor, ensure the config.json snippets are correctly merged into your existing mcpServers section.
• You can extend or customize prompts, tokens, and keywords within the research prompts to tailor results to specific token research needs.
• The tools include robust options for planning (create-research-plan) and for fetching content (fetch-content) which can be stored as resources for reference later. Regularly prune old resources to manage local storage.
• Environment variables for API keys or sources are not required by default, but you may configure additional sources or authentication if your deployment expands beyond public data.

Common issues:

• 403 errors when fetching content: consider using fetch-content with alternative sources or adjust user-agent policies if supported by the source.
• Rate limits: spread requests over time or use search-source/drill-down queries to reduce bulk scraping.
• Configuration mismatches: ensure the mcp.json integration blocks match the host application (Claude Desktop or Cursor) expectations.