Giskard Search is an MCP server designed to give AI agents access to real-time web and news search, with a built-in payment mechanism based on the Lightning Network or USDC on Arbitrum. The project emphasizes "pay-per-use," with no subscription model and no need to register an account.
Main capabilities
- Exposes search capabilities to AI agents via the MCP (Model Context Protocol) protocol.
- Provides real-time web search and news search.
- Supports two payment paths: Lightning Network (paying sats) and USDC on Arbitrum One.
- Includes a get_status() MCP tool for health checks, returning the service name, version, port, uptime, health status, and dependencies.
- Uses DuckDuckGo as the search engine, phoenixd as the Lightning Network node, x402 as the HTTP payment protocol, and Arbitrum as the L2 network for on-chain payment verification.
How to use
- Can be run via Docker.
- Can also be run from source.
- Requires creating a .env file and starting the server.
- The specific installation commands, environment variable configuration, and startup steps are not fully shown in the provided information and are not available for now.
Payment contract information
- Arbitrum One contract address: 0xD467CD1e34515d58F98f8Eb66C0892643ec86AD3
- x402 wallet address: 0xdcc84e9798e8eb1b1b48a31b8f35e5aa7b83dbf4
Use cases
- For AI agents that need real-time information, especially those that need to search autonomously and are willing to pay per use.
- Suitable for developers and agent systems that do not want to use a subscription model or create an account.
Limitations and notes
- The specific pricing, per-query cost, supported payment network details, Docker image name, default port values, environment variable list, source dependencies, and runtime requirements are not specified in the provided information and are not available for now.
- The project belongs to the Mycelium ecosystem; the full scope of that ecosystem's functionality is not expanded on in the provided information and is not available for now.
