-
-
Home page.
-
Agent creation.
-
Interface to explore agents online.
-
Agent specification.
-
Creator payments.
Inspiration
The vision behind btc-mcp is a fully autonomous AI ecosystem, where agents can interact, exchange value, and perform complex tasks without human intervention. Currently, finding and integrating multiple MCP servers or tools into an agent to accomplish specific tasks is a major friction point. btc-mcp simplifies this process by providing agents with easy access to a registry of MCP servers, allowing them to pay for services automatically via the Bitcoin Lightning Network. Creators earn a share of the payments, incentivizing the development of high-quality tools.
What it does
btc-mcp enables creators to easily publish MCP servers powered by Anthropic/Claude to a public registry. Users can access these servers through their own agents by integrating a few tools from a custom Python package into their workflow. Leveraging L402 semantics and the Bitcoin Lightning Network, agents can autonomously pay for server access, allowing them to perform tasks they couldn’t complete before, using the tools provided by creators. Creators earn a share of the earnings and can withdraw anytime on the site.
If users enjoy or find value in a particular MCP server, they can stake some of their credits to help others discover it more easily, creating a community where high-quality servers are recognized and promoted.
Using the available MCP servers is as simple as installing the package and adding tools to your agent. The most simple example is as follows:
from btc_mcp.tools import BTC_MCP_TOOLS, init_lightspark_client
init_lightspark_client(client_id, secret, node_id, node_password)
# Use BTC_MCP_TOOLS with any LangChain-compatible agent
agent = create_react_agent(model=llm, tools=BTC_MCP_TOOLS)
How we built it
Server access and transactions are powered by Lightspark/Bitcoin, including L402-style paywall semantics. Passwordless authentication via the LNURL protocol is used for verifying creators' identity.
The platform uses MCP to standardize agent and tool interfaces, enabling seamless agent-to-agent interactions. Redis manages caching and ephemeral state for Lightning authentication and Docker provides isolated, deployable environments for MCP servers. The Anthropic Python API is used as a middleman, allowing agents to interact with tools seamlessly.
We use vector-based search powered by MongoDB to help agents find servers more intelligently, making discovery faster and more intuitive.
The frontend is built with Next.js for SSR and the backend is built with FastAPI.
Additionally, a Python package was made to allow for anyone to easily integrate btc-mcp functionality into their own agents.
What's next for btc-mcp
While the platform currently supports file transfer, the files are simply encoded in base64 and passed directly into the model's context. In the future, I would like to implement a file handling system rather than embedding the raw data directly in the model input.
Log in or sign up for Devpost to join the conversation.