eac-mcp-commands
The eac-mcp-commands module provides Model Context Protocol (MCP) server integration for AI assistants like Claude. It exposes EAC commands as MCP tools that can be discovered and invoked by AI agents.
System Context
Shows how eac-mcp-commands bridges AI assistants with the EAC system.
Container Architecture
High-level view of the MCP server components.
Component Architecture
Internal structure of the MCP command system.
Workflow Diagrams
Tool Discovery
How AI assistants discover available EAC commands.
Tool Call
The flow when an AI assistant invokes an EAC command.
Design File
- Location:
specs/eac-mcp-commands/.design/workspace.dsl - Interactive:
r2r eac serve-design --module eac-mcp-commands
MCP Integration
The MCP server enables AI assistants to:
- Discover available EAC commands dynamically
- Invoke commands with proper argument handling
- Receive structured responses for further processing
- Access repository state and build information
Configuration
Production Configuration (Default)
MCP server is configured in Claude Code settings (.mcp.json):
Uses r2r eac <command> by default (Docker-based execution).
Development Configuration (Local Override)
For local development with faster startup (.mcp.json):
{
"mcpServers": {
"commands": {
"command": "go",
"args": ["run", "./go/eac/mcp/commands/main.go"],
"env": {
"EAC_USE_DIRECT_BINARY": "true"
}
}
}
}
Uses direct binary execution (faster, ~100ms startup vs ~2s Docker).
Environment Variable
EAC_USE_DIRECT_BINARY:
- true: Forces direct binary execution (development)
- Not set or false: Uses r2r eac (production, default)
Verification
# Test production mode
echo '{"jsonrpc":"2.0","id":1,"method":"initialize"}' | r2r eac mcp commands
# Test development mode
EAC_USE_DIRECT_BINARY=true go run ./go/eac/mcp/commands/main.go < input.json
Tutorials | How-to Guides | Explanation | Reference
You are here: Reference — information-oriented technical descriptions of the system.