Skip to main content
The SDD MCP server exposes the traceability graph to Claude and other MCP clients, enabling real-time queries, impact analysis, and coverage reports without leaving your conversation.

​
What it provides

The server implements the Model Context Protocol specification and offers three categories of capabilities:
  • 5 tools for querying, tracing, and analyzing artifacts
  • Resources via the sdd:// protocol for pipeline state and graph metadata
  • 2 workflow prompts for common multi-step operations

​
Installation

The MCP server is included with the SDD plugin. It registers automatically when you install the plugin in Claude Desktop or any MCP-compatible client.
.mcp.json
{
  "mcpServers": {
    "sdd": {
      "command": "node",
      "args": ["${CLAUDE_PLUGIN_ROOT}/server/dist/index.js"]
    }
  }
}

​
How it works

The server loads the traceability graph from .sdd/traceability-graph.json and keeps it in memory. It builds indexes for fast lookup by artifact ID, type, and relationships. When you call a tool or access a resource:
  1. The server validates the request against the JSON schema
  2. It queries the graph using the prebuilt indexes
  3. It returns structured JSON responses with next-step hints

​
Data source

All data comes from the traceability graph generated by the SDD pipeline stages:
  • Requirements Engineer → REQ artifacts
  • Use Case Engineer → UC artifacts
  • Workflow Engineer → WF artifacts
  • API Design Agent → API artifacts
  • BDD Test Spec Agent → BDD artifacts
  • Task Decomposer → TASK artifacts
  • Code Index → Code/test/commit references (optional)
The graph includes relationships like implements, orchestrates, verifies, and decomposes, plus classification metadata (business domain, technical layer).

​
Next steps

Tools

Query and analyze the graph

Resources

Access pipeline state via sdd:// URIs

Prompts

Run guided workflows

Build docs developers (and LLMs) love

Get started for freeTalk to us