HAL (HTTP API Layer)

HAL (HTTP API Layer)

HAL is a Model Context Protocol (MCP) server that provides HTTP API capabilities to Large Language Models. It allows LLMs to make HTTP requests and interact with web APIs through a secure, controlled interface. HAL can also automatically generate tools from OpenAPI/Swagger specifications for seamless API integration.

Documentation

Complete Documentation →

Visit our comprehensive documentation site for detailed guides, examples, and API reference.

Features

• HTTP GET/POST/PUT/PATCH/DELETE/OPTIONS/HEAD Requests: Fetch and send data to any HTTP endpoint
• Secure Secret Management: Environment-based secrets with {secrets.key} substitution and automatic redaction
• Swagger/OpenAPI Integration: Automatically generate tools from API specifications
• Built-in Documentation: Self-documenting API reference
• Secure: Runs in isolated environment with controlled access
• Fast: Built with TypeScript and optimized for performance

Usage

HAL is designed to work with MCP-compatible clients. Here are some examples:

Basic Usage (Claude Desktop)

Add HAL to your Claude Desktop configuration (npx will automatically install and run HAL):

With Swagger/OpenAPI Integration and Secrets

To enable automatic tool generation from an OpenAPI specification and use secrets:

Direct Usage

Configuration

HAL supports the following environment variables:

• HAL_SWAGGER_FILE: Path to OpenAPI/Swagger specification file (JSON or YAML format)
• HAL_API_BASE_URL: Base URL for API requests (overrides the servers specified in the OpenAPI spec)
• HAL_SECRET_*: Secret values for secure substitution in requests (e.g., HAL_SECRET_TOKEN=abc123)
• HAL_ALLOW_*: URL restrictions for namespaced secrets (e.g., HAL_ALLOW_MICROSOFT="https://azure.microsoft.com/*")
• HAL_WHITELIST_URLS: Comma-separated list of URL patterns that are allowed (if set, only these URLs are permitted)
• HAL_BLACKLIST_URLS: Comma-separated list of URL patterns that are blocked (if set, these URLs are denied)

Secret Management

HAL provides secure secret management to keep sensitive information like API keys, tokens, and passwords out of the conversation while still allowing the AI to use them in HTTP requests.

How It Works

1. Environment Variables: Define secrets using the HAL_SECRET_ prefix:
   HAL_SECRET_API_KEY=your-secret-api-key HAL_SECRET_TOKEN=your-auth-token HAL_SECRET_USERNAME=your-username
2. Template Substitution: Reference secrets in your requests using {secrets.key} syntax:
   • URLs: https://api.example.com/data?token={secrets.token}
   • Headers: {"Authorization": "Bearer {secrets.api_key}"}
   • Request Bodies: {"username": "{secrets.username}", "password": "{secrets.password}"}
3. Security: The AI never sees the actual secret values, only the template placeholders. Values are substituted at request time.

Automatic Secret Redaction

HAL automatically redacts secret values from all responses sent back to the AI, providing an additional layer of security against credential exposure.

How It Works

1. Secret Tracking: HAL maintains a registry of all secret values from environment variables
2. Response Scanning: All HTTP responses (headers, bodies, error messages) are scanned for secret values
3. Automatic Replacement: Any occurrence of actual secret values is replaced with [REDACTED] before sending to the AI
4. Comprehensive Coverage: Redaction applies to:
   • Error messages (including URL parsing errors that might expose credentials)
   • Response headers (in case APIs echo back authentication data)
   • Response bodies (protecting against API responses that might include sensitive data)
   • All other text returned to the AI

Example Protection

Before (vulnerable):

After (secure):

This protection is automatic and requires no configuration - HAL will redact any secret values regardless of how they appear in responses, ensuring that even if an API or error message attempts to expose credentials, the AI never sees the actual values.

Namespaces and URL Restrictions

HAL supports organizing secrets into namespaces and restricting them to specific URLs for enhanced security:

Namespace Convention

Use - for namespace separators and _ for word separators within keys:

URL Restrictions

Restrict namespaced secrets to specific URLs using HAL_ALLOW_* environment variables:

How Parsing Works

Understanding how environment variable names become template keys:

Step-by-step breakdown:

1. Remove HAL_SECRET_ prefix → AZURE-STORAGE_ACCESS_KEY
2. Split on first _ → Namespace: AZURE-STORAGE, Key: ACCESS_KEY
3. Transform namespace: AZURE-STORAGE → azure.storage (dashes become dots, lowercase)
4. Transform key: ACCESS_KEY → access_key (underscores stay, lowercase)
5. Combine: {secrets.azure.storage.access_key}

More Examples

Visual Guide: Complete Flow

Security Benefits

• Principle of Least Privilege: Secrets only work with their intended services
• Prevents Cross-Service Leakage: Azure secrets can't be sent to AWS APIs
• Defense in Depth: Even with AI errors or prompt injection, secrets are constrained
• Clear Organization: Namespace structure makes secret management more intuitive

Real-World Usage Scenarios

Scenario 1: Multi-Cloud Application

Usage in requests:

✅ Works: URL matches Azure Storage pattern
❌ Blocked: If used with https://s3.amazonaws.com/bucket - wrong service!

Scenario 2: Development vs Production

Scenario 3: Department Isolation

Error Examples

When URL restrictions are violated, you get clear error messages:

This helps you quickly identify:

• Which secret was blocked
• What URL was attempted
• What URLs are actually allowed

Quick Reference

Pattern: HAL_SECRET_<NAMESPACE>_<KEY> → {secrets.<namespace>.<key>} + HAL_ALLOW_<NAMESPACE>

Backward Compatibility

Non-namespaced secrets (without URL restrictions) continue to work as before:

URL Filtering

HAL supports global URL filtering to control which URLs can be accessed through whitelist or blacklist patterns. This provides an additional security layer beyond the namespace-based secret restrictions.

Whitelist Mode

When HAL_WHITELIST_URLS is set, only URLs matching the specified patterns are allowed:

Blacklist Mode

When HAL_BLACKLIST_URLS is set, all URLs are allowed except those matching the specified patterns:

Pattern Syntax

URL patterns support wildcard matching using *:

• https://api.example.com/* - Matches any path under the API
• https://*.example.com/* - Matches any subdomain
• *://internal.company.com/* - Matches any protocol

Important Notes

• Whitelist takes precedence: If both HAL_WHITELIST_URLS and HAL_BLACKLIST_URLS are set, the whitelist is used and a warning is logged
• Global filtering: This applies to all HTTP requests, regardless of secrets or tools used
• Case-insensitive: URL pattern matching is case-insensitive
• No filtering by default: If neither environment variable is set, all URLs are allowed

Examples

Example Usage

The {secrets.github_token} will be replaced with the value of HAL_SECRET_GITHUB_TOKEN environment variable before making the request.

Available Tools

Built-in HTTP Tools

These tools are always available regardless of configuration:

list-secrets

Get a list of available secret keys that can be used with {secrets.key} syntax.

Parameters: None

Example Response:

Security Note: Only shows the key names, never the actual secret values.

http-get

Make HTTP GET requests to any URL.

Parameters:

• url (string, required): The URL to request
• headers (object, optional): Additional headers to send

Example:

http-post

Make HTTP POST requests with optional body and headers.

Parameters:

• url (string, required): The URL to request
• body (string, optional): Request body content
• headers (object, optional): Additional headers to send
• contentType (string, optional): Content-Type header (default: "application/json")

Example:

Auto-generated Swagger/OpenAPI Tools

When you provide a Swagger/OpenAPI specification via HAL_SWAGGER_FILE, HAL will automatically generate tools for each endpoint defined in the specification. These tools are named using the pattern swagger_{operationId} and include:

• Automatic parameter validation based on the OpenAPI schema
• Path parameter substitution (e.g., /users/{id} → /users/123)
• Query parameter handling
• Request body support for POST/PUT/PATCH operations
• Proper HTTP method mapping

For example, if your OpenAPI spec defines an operation with operationId: "getUser", HAL will create a tool called swagger_getUser that you can use directly.

Available Resources

docs://hal/api

Access comprehensive API documentation and usage examples, including documentation for any auto-generated Swagger tools.

OpenAPI/Swagger Integration Details

Supported OpenAPI Features

• ✅ OpenAPI 3.x and Swagger 2.x specifications
• ✅ JSON and YAML format support
• ✅ Path parameters (/users/{id})
• ✅ Query parameters
• ✅ Request body (JSON, form-encoded)
• ✅ All HTTP methods (GET, POST, PUT, PATCH, DELETE, etc.)
• ✅ Parameter validation (string, number, boolean, arrays)
• ✅ Required/optional parameter handling
• ✅ Custom headers support

Example OpenAPI Integration

Given this OpenAPI specification:

HAL will automatically create a swagger_getUser tool that the LLM can use like:

This will make a GET request to https://api.example.com/v1/users/123.

Development

Prerequisites

• Node.js 18 or later
• npm or yarn

Setup

Scripts

• npm run build - Build the TypeScript project
• npm run dev - Run in development mode with hot reload
• npm start - Start the built server
• npm run lint - Run ESLint
• npm test - Run tests

Security Considerations

• HAL makes actual HTTP requests to external services
• Use appropriate authentication and authorization for your APIs
• Be mindful of rate limits and API quotas
• Consider network security and firewall rules
• When using Swagger integration, ensure your OpenAPI specifications are from trusted sources

Contributing

1. Fork the repository
2. Create a feature branch (git checkout -b feature/amazing-feature)
3. Commit your changes (git commit -m 'Add some amazing feature')
4. Push to the branch (git push origin feature/amazing-feature)
5. Open a Pull Request

License

This project is licensed under the MIT License - see the LICENSE file for details.

Acknowledgments

• Built with the Model Context Protocol TypeScript SDK
• Inspired by the need for LLMs to interact with web APIs safely and efficiently
• OpenAPI integration powered by swagger-parser