MCP 服务器

Darwin Standards MCP Server

Provides standards documentation and validation tools for the Darwin platform, including naming conventions and agent card validation. It enables AI agents to search platform standards and access resource templates for design and implementation.

README

Darwin Standards MCP Server

MCP server providing standards documentation and validation tools for the Darwin platform.

Overview

The Standards MCP Server is designed to:

Expose standards as MCP resources - Agents can read platform standards via standard URIs
Provide validation tools - Validate naming conventions, tool definitions, and agent cards
Enable standards search - Find relevant standards documentation quickly

Note: This is an MCP Server, not a true Agent. It provides tools and resources but does not contain an LLM and does not make autonomous decisions.

Quick Start

Option 1: Docker (Recommended for local development)

# Build and run
make build
make run

# Or with docker directly
docker build -t standards-mcp-server .
docker run -p 8080:8080 \
  -v /path/to/standards:/app/standards:ro \
  -e MCP_TRANSPORT=http \
  standards-mcp-server

Option 2: Local Installation

# Create virtual environment and install
make install

# Run locally (stdio mode)
make run-local

# Run locally (HTTP mode)
make run-http

Option 3: Claude Code Integration

# Using Python module directly
claude mcp add --transport stdio \
  --env STANDARDS_PATH=/path/to/docs/standards \
  standards -- python -m standards_mcp_server.server

# Using container
claude mcp add --transport stdio standards -- \
  docker run --rm -i \
  -v /path/to/docs/standards:/app/standards:ro \
  -e MCP_TRANSPORT=stdio \
  fgsansharedacrdeveus.azurecr.io/standards-mcp-server:latest

MCP Resources

Standards Documentation

URI	Description
`standards://agents/design`	Agent design principles
`standards://agents/communication`	A2A protocol standards
`standards://agents/tools`	MCP tool integration
`standards://agents/memory`	Neo4j memory architecture
`standards://agents/observability`	OpenTelemetry setup
`standards://agents/security`	Security requirements
`standards://mcp/server-design`	MCP server architecture
`standards://mcp/tool-design`	Tool definition standards
`standards://mcp/resource-design`	Resource exposure standards
`standards://mcp/security`	MCP security patterns
`standards://mcp/error-handling`	Error handling standards

Templates and Implementations

URI	Description
`templates://mcp/python`	Python MCP server template
`templates://mcp/dotnet`	.NET MCP server template
`implementations://mcp/standards`	This server's implementation
`implementations://agents/python/langchain`	LangChain agent

MCP Tools

Validation Tools

validate_azure_resource_name - Validate Azure resource names against naming conventions
validate_mcp_tool_definition - Validate MCP tool definitions against standards
validate_agent_card - Validate A2A agent cards against protocol standards

Search Tools

search_standards - Full-text search across standards documents
list_standards_categories - List all available categories
get_standards_summary - Get document summaries for a category

Configuration

Variable	Description	Default
`MCP_SERVER_NAME`	Server name	`standards-mcp-server`
`MCP_TRANSPORT`	Transport type (`stdio` or `http`)	`stdio`
`MCP_PORT`	HTTP port	`8080`
`MCP_HOST`	HTTP host	`0.0.0.0`
`MCP_LOG_LEVEL`	Log level	`INFO`
`STANDARDS_PATH`	Path to standards docs	`docs/standards`
`REFERENCE_IMPLEMENTATIONS_PATH`	Path to implementations	`reference-implementations`
`OTEL_ENABLED`	Enable OpenTelemetry	`true`
`OTEL_EXPORTER_OTLP_ENDPOINT`	OTLP endpoint	`null`

Development

Running Tests

# Run all tests
make test

# With coverage
make test-cov

# Linting
make lint

# Type checking
make typecheck

Project Structure

darwin-standards-mcp/
├── .github/
│   ├── workflows/          # CI/CD pipelines
│   └── PULL_REQUEST_TEMPLATE.md
├── .k8s/
│   ├── base/              # Base Kubernetes manifests
│   └── overlays/          # Environment-specific overlays
│       ├── dev/
│       └── prod/
├── src/standards_mcp_server/
│   ├── __init__.py
│   ├── server.py          # Main MCP server
│   ├── config.py          # Configuration
│   ├── observability.py   # OpenTelemetry setup
│   ├── resources/         # MCP resources
│   └── tools/             # MCP tools
├── tests/
├── scripts/               # Deployment scripts
├── docs/                  # Documentation
├── Dockerfile
├── Makefile
└── pyproject.toml

Deployment

Deploy to AKS

# Using make
make deploy

# Using script
./scripts/deploy-to-aks.sh

Manual Deployment

# Get AKS credentials
az aks get-credentials --resource-group rg-fgsan-shared-dev-eus --name fgsan-shared-aks-dev-eus

# Apply manifests
kubectl apply -k .k8s/overlays/dev

# Verify
kubectl get pods -n darwin-mcp -l app=standards-mcp-server

See DEPLOYMENT.md for detailed deployment instructions.

CI/CD

The repository includes GitHub Actions workflows for:

lint-and-test.yml - Runs on PRs: linting, testing, type checking
build-and-push.yml - Runs on push to main: builds and pushes Docker image to ACR
deploy-to-aks.yml - Runs after build: deploys to AKS
release.yml - Runs on version tags: creates GitHub releases