TurboMCP Macros

Zero-boilerplate procedural macros for ergonomic MCP server development with automatic schema generation and compile-time validation.

Overview

turbomcp-macros provides the procedural macros that make TurboMCP development effortless. These macros eliminate boilerplate code while providing compile-time validation, automatic schema generation, and type-safe parameter handling.

Key Features

🎯 Zero Boilerplate

• Automatic registration - Tools and resources registered automatically
• Schema generation - JSON schemas generated from Rust types
• Parameter extraction - Type-safe parameter conversion and validation
• Error handling - Automatic error type conversion and propagation

✅ Compile-Time Validation

• Type checking - Parameter types validated at compile time
• Schema validation - Generated schemas validated for correctness
• IDE support - Full IntelliSense and error reporting
• Macro hygiene - Proper variable scoping and name collision prevention

📋 Automatic Schema Generation

• JSON Schema - Complete JSON Schema generation from Rust types
• Parameter documentation - Extract documentation from function signatures
• Type introspection - Deep analysis of parameter and return types
• Schema caching - Efficient schema generation and reuse

🔍 Context Injection

• Flexible positioning - Context parameter can appear anywhere in function signature
• Send-safe - Proper Send/Sync bounds for async context
• Type safety - Compile-time validation of context usage
• Optional context - Functions can opt-in or out of context injection

Architecture

┌─────────────────────────────────────────────┐
│              TurboMCP Macros                │
├─────────────────────────────────────────────┤
│ Procedural Macro Processing                │
│ ├── #[server] trait implementation         │
│ ├── #[tool] function registration          │
│ ├── #[resource] handler registration       │
│ └── #[prompt] template registration        │
├─────────────────────────────────────────────┤
│ Schema Generation Engine                   │
│ ├── Type introspection                     │
│ ├── JSON Schema creation                   │
│ ├── Parameter validation                   │
│ └── Documentation extraction               │
├─────────────────────────────────────────────┤
│ Code Generation                            │
│ ├── Handler registration code              │
│ ├── Parameter extraction logic             │
│ ├── Error conversion helpers               │
│ └── Schema metadata functions              │
├─────────────────────────────────────────────┤
│ Compile-Time Validation                    │
│ ├── Type compatibility checking            │
│ ├── Parameter validation                   │
│ ├── Context injection validation           │
│ └── Schema correctness verification        │
└─────────────────────────────────────────────┘

Core Macros

#[server] - Server Implementation

Automatically implements the MCP server trait for a struct:

use turbomcp::prelude::*;

#[derive(Clone)]
struct Calculator;

#[server]
impl Calculator {
    #[tool("Add two numbers")]
    async fn add(&self, a: f64, b: f64) -> McpResult<f64> {
        Ok(a + b)
    }
    
    #[tool("Get server status")]
    async fn status(&self, ctx: Context) -> McpResult<String> {
        ctx.info("Status requested").await?;
        Ok("Server running".to_string())
    }
}

// Generated code includes:
// - Automatic trait implementation
// - Handler registration
// - Schema generation
// - Transport integration

Generated Capabilities:

• Automatic MCP trait implementation
• Handler registry setup with all annotated functions
• Schema generation for all tools/resources/prompts
• Transport method implementations (run_stdio, run_http, etc.)

#[tool] - Tool Registration

Transforms functions into MCP tools with automatic parameter handling:

#[tool("Calculate mathematical expressions")]
async fn calculate(
    #[description("Mathematical expression to evaluate")]
    expression: String,
    #[description("Precision for floating point results")]
    precision: Option<u32>,
    ctx: Context
) -> McpResult<serde_json::Value> {
    ctx.info(&format!("Calculating: {}", expression)).await?;
    
    let precision = precision.unwrap_or(2);
    // ... calculation logic
    
    Ok(serde_json::json!({
        "result": result,
        "expression": expression,
        "precision": precision
    }))
}

Generated Features:

• JSON Schema with parameter descriptions
• Type-safe parameter extraction from JSON
• Optional parameter handling
• Context injection (can appear anywhere in signature)
• Automatic error conversion
• Tool metadata functions

#[resource] - Resource Registration

Creates URI template-based resource handlers:

#[resource("file://{path}")]
async fn read_file(
    #[description("File path to read")]
    path: String,
    #[description("Maximum file size in bytes")]
    max_size: Option<usize>,
    ctx: Context
) -> McpResult<String> {
    let max_size = max_size.unwrap_or(1024 * 1024); // 1MB default
    
    if std::fs::metadata(&path)?.len() > max_size as u64 {
        return Err(McpError::InvalidInput("File too large".to_string()));
    }
    
    ctx.info(&format!("Reading file: {}", path)).await?;
    
    tokio::fs::read_to_string(&path).await
        .map_err(|e| McpError::Resource(e.to_string()))
}

URI Template Features:

• Automatic URI pattern matching
• Path parameter extraction
• Query parameter support
• URI validation
• Resource metadata generation

#[prompt] - Prompt Template Registration

Creates prompt templates with parameter substitution:

#[prompt("code_review")]
async fn code_review_prompt(
    #[description("Programming language")]
    language: String,
    #[description("Code to review")]
    code: String,
    #[description("Focus areas for review")]
    focus: Option<Vec<String>>,
    ctx: Context
) -> McpResult<String> {
    let focus_areas = focus.unwrap_or_else(|| vec![
        "security".to_string(),
        "performance".to_string(),
        "maintainability".to_string()
    ]);
    
    ctx.info(&format!("Generating {} code review prompt", language)).await?;
    
    Ok(format!(
        "Please review the following {} code focusing on {}:\n\n```{}\n{}\n```",
        language,
        focus_areas.join(", "),
        language,
        code
    ))
}

Advanced Features

Context Injection

The Context parameter can appear anywhere in the function signature:

// Context first
#[tool("Process data")]
async fn process(ctx: Context, data: String) -> McpResult<String> {
    ctx.info("Processing started").await?;
    Ok(format!("Processed: {}", data))
}

// Context in middle
#[tool("Transform data")]
async fn transform(input: String, ctx: Context, format: String) -> McpResult<String> {
    ctx.info(&format!("Transforming to {}", format)).await?;
    // transformation logic
    Ok(transformed)
}

// Context last
#[tool("Validate input")]
async fn validate(data: String, strict: bool, ctx: Context) -> McpResult<bool> {
    ctx.info("Validating input").await?;
    // validation logic
    Ok(is_valid)
}

// No context
#[tool("Simple calculation")]
async fn add(a: f64, b: f64) -> McpResult<f64> {
    Ok(a + b)
}

Parameter Descriptions

Use the #[description] attribute for rich parameter documentation:

#[tool("Search documents")]
async fn search(
    #[description("Search query string")]
    query: String,
    
    #[description("Maximum number of results to return")]
    #[default(10)]
    limit: Option<u32>,
    
    #[description("Include archived documents in search")]
    #[default(false)]
    include_archived: Option<bool>,
    
    #[description("Sort results by relevance or date")]
    #[allowed("relevance", "date")]
    sort_by: Option<String>,
) -> McpResult<SearchResults> {
    // Implementation
}

Generated Schema:

{
  "type": "object",
  "properties": {
    "query": {
      "type": "string",
      "description": "Search query string"
    },
    "limit": {
      "type": "integer",
      "description": "Maximum number of results to return",
      "default": 10
    },
    "include_archived": {
      "type": "boolean", 
      "description": "Include archived documents in search",
      "default": false
    },
    "sort_by": {
      "type": "string",
      "description": "Sort results by relevance or date",
      "enum": ["relevance", "date"]
    }
  },
  "required": ["query"]
}

Custom Types and Schema Generation

The macros automatically generate schemas for custom types:

#[derive(Serialize, Deserialize)]
struct User {
    id: u64,
    name: String,
    email: Option<String>,
    active: bool,
}

#[derive(Serialize, Deserialize)]
struct CreateUserRequest {
    name: String,
    email: String,
    role: UserRole,
}

#[derive(Serialize, Deserialize)]
enum UserRole {
    Admin,
    User,
    Guest,
}

#[tool("Create a new user")]
async fn create_user(request: CreateUserRequest) -> McpResult<User> {
    // Schema automatically generated for both CreateUserRequest and User
    // Enums become string unions in JSON Schema
    // Optional fields marked appropriately
    Ok(User {
        id: generate_id(),
        name: request.name,
        email: Some(request.email),
        active: true,
    })
}

Error Handling Macros

Ergonomic error creation macros:

use turbomcp::prelude::*;

#[tool("Divide numbers")]
async fn divide(a: f64, b: f64) -> McpResult<f64> {
    if b == 0.0 {
        return Err(mcp_error!("Division by zero: {} / {}", a, b));
    }
    
    Ok(a / b)
}

#[tool("Process file")]
async fn process_file(path: String) -> McpResult<String> {
    let content = tokio::fs::read_to_string(&path).await
        .map_err(|e| mcp_error!("Failed to read file {}: {}", path, e))?;
    
    // Processing logic
    Ok(processed_content)
}

Metadata Access

The macros generate metadata access functions:

#[derive(Clone)]
struct MyServer;

#[server]
impl MyServer {
    #[tool("Example tool")]
    async fn example(&self, input: String) -> McpResult<String> {
        Ok(input)
    }
}

// Generated metadata functions
let (name, description, schema) = MyServer::example_tool_metadata();
assert_eq!(name, "example");
assert_eq!(description, "Example tool");
// schema contains the complete JSON Schema

// Test the generated function directly
let result = MyServer.test_tool_call("example", serde_json::json!({
    "input": "test"
})).await?;

Macro Attributes

Tool Attributes

Resource Attributes

Generated Code Examples

Tool Registration

Input:

#[tool("Add numbers")]
async fn add(&self, a: f64, b: f64) -> McpResult<f64> {
    Ok(a + b)
}

Generated (simplified):

// Metadata function
pub fn add_tool_metadata() -> (&'static str, &'static str, serde_json::Value) {
    ("add", "Add numbers", serde_json::json!({
        "type": "object",
        "properties": {
            "a": {"type": "number"},
            "b": {"type": "number"}
        },
        "required": ["a", "b"]
    }))
}

// Handler registration
async fn register_handlers(&self, registry: &mut HandlerRegistry) -> McpResult<()> {
    registry.register_tool("add", |params| {
        let a: f64 = extract_param(&params, "a")?;
        let b: f64 = extract_param(&params, "b")?;
        self.add(a, b).await
    }).await?;
    
    Ok(())
}

// Direct test function
pub async fn test_tool_call(&self, name: &str, params: serde_json::Value) -> McpResult<serde_json::Value> {
    match name {
        "add" => {
            let a: f64 = extract_param(&params, "a")?;
            let b: f64 = extract_param(&params, "b")?;
            let result = self.add(a, b).await?;
            Ok(serde_json::to_value(result)?)
        },
        _ => Err(McpError::InvalidInput(format!("Unknown tool: {}", name)))
    }
}

IDE Integration

The macros provide excellent IDE support:

• IntelliSense - Full auto-completion for generated functions
• Error highlighting - Compile-time error detection
• Type information - Hover information for generated code
• Go to definition - Navigate to macro-generated implementations
• Refactoring support - Safe renaming and extraction

Testing Support

The macros generate testing utilities:

#[cfg(test)]
mod tests {
    use super::*;
    
    #[tokio::test]
    async fn test_calculator_tools() {
        let calc = Calculator;
        
        // Test add tool directly
        let result = calc.test_tool_call("add", serde_json::json!({
            "a": 5.0,
            "b": 3.0
        })).await.unwrap();
        
        assert_eq!(result, serde_json::json!(8.0));
        
        // Test schema generation
        let (name, desc, schema) = Calculator::add_tool_metadata();
        assert_eq!(name, "add");
        assert_eq!(desc, "Add two numbers");
        assert!(schema["properties"]["a"]["type"] == "number");
    }
}

Performance

The macros generate efficient code:

• Zero runtime overhead - All processing happens at compile time
• Optimized registration - Efficient handler lookup and dispatch
• Schema caching - Schemas generated once and reused
• Minimal allocations - Smart parameter extraction with minimal copying

Development

Building

# Build macros crate
cargo build

# Test macro expansion
cargo expand --package turbomcp-macros

# Run macro tests
cargo test

Debugging Macros

# See expanded macro code
cargo expand --bin my_server

# Debug specific macro
RUST_LOG=debug cargo build

Related Crates

• turbomcp - Main framework (uses these macros)
• turbomcp-core - Core types and utilities
• turbomcp-server - Server framework
• turbomcp-protocol - MCP protocol implementation

External Resources

• The Rust Reference - Procedural Macros - Rust macro documentation
• JSON Schema Specification - Schema format specification
• serde Documentation - Serialization framework used by macros

License

Licensed under the MIT License.

Part of the TurboMCP high-performance Rust SDK for the Model Context Protocol.