solscan-mcp is a Model Context Protocol (MCP) server written in Rust, designed to interact with the Solscan Pro API and provide a natural language interface for querying Solana blockchain data. It enables large language models to query token information, account activity, and transaction details on the Solana blockchain in natural language.
Main capabilities: The server provides tools for querying token information, account activity, and transaction details, supporting simple prompts and complex queries. For example, a user can upload a CSV list of suspected addresses and, combined with the perplexity-mcp research tool, have the LLM supplement in context information about how investigators define criminal wallets based on activity (such as MEV, dusting attacks, poisoning, sandwich attacks, etc.), and then solscan-mcp uses this context to investigate the wallets and provide a report. Detailed documentation for all available tools is in TOOLS.md.
Usage: Before use, a Rust toolchain (installed via rustup) is required for regular use, Docker for running via Docker, and a Solscan Pro API key (available from Solscan APIs). During configuration, the relevant settings need to be added to claude_desktop_config.json or claude_config.json. The specific configuration content is currently marked as WIP and will be provided soon.
Applicable scenarios and limitations: It is suitable for scenarios that require querying Solana blockchain data in natural language, especially for complex analysis combined with multiple MCPs, such as investigating suspicious wallet activity. The project is positioned as a practical script. Considering that the structure of the APIs it depends on may change, the code organization allows api.rs to be split according to Solscan namespaces. The design goal focuses on usability and ease of maintenance, and ensures that errors are gracefully ignored without interrupting the flow (unlike some Python/TS MCPs that crash on simple errors). More details such as fees, pricing, and platforms are not yet provided.
