bsl-agent-mcp
MCP server that lets AI agents operate a user-owned wallet on the Bitcoin-IPC testnet, enabling BTC deposits, cross-subnet wBTC payments, and anchored payment receipts, with local key signing and spend caps.
README
bsl-agent-mcp
MCP server that lets an AI agent (Claude Desktop, Claude Code, or any MCP
client) operate a user-owned wallet on the Bitcoin-IPC testnet: deposit L1 BTC
into subnet1, pay wBTC within subnet1 or across to subnet2, and fetch the
Bitcoin-anchored receipt for any payment. The agent never holds keys — it can
only request operations; the server signs locally and refuses any deposit or
payment above SPEND_CAP_SATS until the user explicitly approves it
(confirmed=true).
Disclaimer: this code is not audited and is meant for testnet use only. Do not use it with mainnet funds or keys that control real value.
Tools
| Tool | What it does |
|---|---|
wallet_status |
L1 BTC balance + wBTC balances on subnet1/subnet2 + spending cap |
deposit_btc |
L1 BTC → wBTC at the agent's address on subnet1 |
pay |
wBTC to an address-book payee or user-supplied address; same-subnet (seconds) or subnet1→subnet2 (next checkpoint) |
address_book |
Known payees with address, subnet, and live wBTC balance |
check_balance |
wBTC balance of any account (payee or raw address) on both subnets |
payment_receipt |
The checkpoint anchoring a payment: Bitcoin block height + txid |
subnet_info |
Committee, threshold, multisig address, latest checkpoint of a subnet |
Requirements
A running bsl-client container — the public image is
ghcr.io/bitcoinscalinglabs/bsl-client:testnet — with its bsl.env
configuration file, which must also contain AGENT_PRIVATE_KEY (the EVM key
the server pays from).
Run with Docker
The image runs next to bsl-client and shares its network; the entrypoint
generates the ipc-cli config, imports AGENT_PRIVATE_KEY, runs connectivity
checks, and idles.
docker build -t ghcr.io/bitcoinscalinglabs/bsl-agent-mcp:testnet .
docker run -d --env-file ./bsl.env \
--network container:bsl-client \
--name bsl-agent-mcp \
ghcr.io/bitcoinscalinglabs/bsl-agent-mcp:testnet
docker logs bsl-agent-mcp # should end with "bsl-agent-mcp ready"
Claude Desktop attaches MCP sessions to the standing container:
{
"mcpServers": {
"bsl-payments": {
"command": "docker",
"args": ["exec", "-i", "bsl-agent-mcp", "node", "/app/dist/index.js"]
}
}
}
Run from source (development)
Node.js ≥ 20. The bsl-client container must publish the provider port
(-p 127.0.0.1:3030:3030) and have the agent key imported once:
docker exec bsl-client ipc-cli wallet import --wallet-type evm --private-key <key>
npm install && npm run build
cp .env.example .env # fill in
Claude Desktop config: "command": "node",
"args": ["/ABS/PATH/bsl-agent-mcp/dist/index.js"]. Configuration comes from
the repo's .env; an env block in the Desktop config takes precedence.
Configuration
| Var | Required | Meaning |
|---|---|---|
API_TOKEN (or TOKEN) |
yes | Access key for the L2 gateways and the bsl-client provider |
AGENT_PRIVATE_KEY |
yes | EVM key the server pays from (holds wBTC) |
SUBNET_ID / SUBNET2_ID |
yes | Subnet ids (SUBNET_ID2 also accepted) |
URL / URL2 |
no | Gateway RPC endpoints (default: BSL testnet gateways) |
PROVIDER_URL |
no | bsl-client provider (default http://127.0.0.1:3030/api) |
BSL_CONTAINER |
no | Container name for docker exec in from-source mode (default bsl-client) |
SPEND_CAP_SATS |
no | Per-operation cap; above it, user approval required (default 500000) |
Notes
- Amounts are satoshis everywhere (1 BTC = 100,000,000 sats); wBTC has 18 decimals and the server converts internally.
- Recipients are either payees from the built-in address book or addresses supplied by the user; the agent is instructed never to invent an address.
Recommended Servers
playwright-mcp
A Model Context Protocol server that enables LLMs to interact with web pages through structured accessibility snapshots without requiring vision models or screenshots.
Magic Component Platform (MCP)
An AI-powered tool that generates modern UI components from natural language descriptions, integrating with popular IDEs to streamline UI development workflow.
Audiense Insights MCP Server
Enables interaction with Audiense Insights accounts via the Model Context Protocol, facilitating the extraction and analysis of marketing insights and audience data including demographics, behavior, and influencer engagement.
VeyraX MCP
Single MCP tool to connect all your favorite tools: Gmail, Calendar and 40 more.
graphlit-mcp-server
The Model Context Protocol (MCP) Server enables integration between MCP clients and the Graphlit service. Ingest anything from Slack to Gmail to podcast feeds, in addition to web crawling, into a Graphlit project - and then retrieve relevant contents from the MCP client.
Kagi MCP Server
An MCP server that integrates Kagi search capabilities with Claude AI, enabling Claude to perform real-time web searches when answering questions that require up-to-date information.
E2B
Using MCP to run code via e2b.
Neon Database
MCP server for interacting with Neon Management API and databases
Exa Search
A Model Context Protocol (MCP) server lets AI assistants like Claude use the Exa AI Search API for web searches. This setup allows AI models to get real-time web information in a safe and controlled way.
Qdrant Server
This repository is an example of how to create a MCP server for Qdrant, a vector search engine.