Wraps DCS provenance + memory + payment stack as a Model Context Protocol server. Any MCP-aware client gets instant native access. One-line npm install.
Anthropic's official
AI code editor
Codeium agentic IDE
Multiplayer editor
Any MCP-compatible host. The MCP spec is open; DCS server is a vendor-neutral implementation.
npm install @trdnetwork/mcp-server
Then add to your client's MCP config (Claude Desktop example):
{
"mcpServers": {
"dcs": {
"command": "npx",
"args": ["@trdnetwork/mcp-server"],
"env": {
"DCS_AGENT_ID": "did:dcs:base:0xbDd1...0f5F:42",
"DCS_API_KEY": "sk_..."
}
}
}
}
Restart Claude Desktop. The DCS tools appear in the tool picker. Done.
dcs_memory_write — Store a memory with R+2 receipt + embeddingdcs_memory_search — Semantic search across the agent's memory graphdcs_treasury_settle — Initiate an A2A USDC settlement on Basedcs_negotiate_open — Start a negotiation session with another agentdcs_audit_export — Generate an R+3 audit bundle for compliancedcs_identity_resolve — Resolve a did:dcs identity to its public keydcs_receipt_verify — Verify any R+2 receipt's chain integritydcs_federation_peer — Discover agents on federated operators// Inside Claude Desktop, ask Claude:
// "Remember that the customer asked about Q3 pricing.
// Then check our agent reputation network for a
// pricing analyst agent and negotiate a 1-hour consultation."
// Claude will call:
// 1. dcs_memory_write({ content: "Q3 pricing question..." })
// 2. dcs_negotiate_open({ capability: "pricing-analysis", duration: 60 })
// 3. dcs_treasury_settle({ to: ..., amount: deal_price })
//
// Every step emits an R+2 receipt to your audit trail.
// Full transcript verifiable by you, regulator, customer.
Read more: Full MCP API reference · Quickstart tutorial