HTTP client tool used for testing inference endpoints and API connectivity
Containerized deployment option for running the MCP server in production environments
Environment variable management for secure API key and configuration handling
Version control integration for managing MCP server source code and installations
Source code hosting and distribution platform for installing the MCP server directly from repositories
Supported platform for Claude Desktop configuration and MCP server deployment
Supported platform for Claude Desktop configuration and MCP server deployment
Documentation format support for generating tool documentation and reports
Uses Pydantic models for full validation of all inputs and outputs to ensure type safety across API interactions
Provides testing framework for development and validation of MCP server functionality
Python-based MCP server implementation for integrating with the Xplainable AI platform
Code linting and formatting tool used in the development workflow
Xplainable MCP Server
A Model Context Protocol (MCP) server that provides secure access to Xplainable AI platform capabilities through standardized tools and resources.
Features
- Secure Authentication: Token-based authentication with environment variable management
- Read Operations: Access models, deployments, preprocessors, and collections
- Write Operations: Deploy models, manage deployments, generate reports (with proper authorization)
- Type Safety: Full Pydantic model validation for all inputs/outputs
- Rate Limiting: Built-in rate limiting and request validation
- Audit Logging: Comprehensive logging of all operations
Installation
CLI Commands
The server includes a CLI for management and documentation:
Quick Start
For Production Users
If you just want to use this MCP server with Claude Code:
- Get your Xplainable API key from https://platform.xplainable.io
- Add the MCP configuration (see Claude Code Configuration above)
- That's it! Claude Code will handle installation automatically
For Developers
1. Set up environment variables
Create a .env
file with your Xplainable credentials:
2. Run the server
3. Connect with an MCP client
Claude Code Configuration
Option 1: Install from GitHub (Recommended)
Option 2: Clone and run from source
Option 3: Development with local backend
Claude Desktop Configuration
Add the configuration to your Claude Desktop MCP settings file:
File Locations:
- macOS:
~/Library/Application Support/Claude/claude_desktop_config.json
- Windows:
%APPDATA%\Claude\claude_desktop_config.json
- Linux:
~/.config/Claude/claude_desktop_config.json
Option 1: Install from GitHub (Recommended)
Option 2: Development setup (from source)
Option 3: Using conda environment
Development Setup
For Local Development with Claude Code
- Set up the environment:
- Configure environment variables:
- Test the setup:
Example Deployment Workflow
Here's a complete example of deploying a model and testing inference:
Available Tools
Discovery Tools
list_tools()
- List all available MCP tools with descriptions and parameters
Read-Only Tools
get_connection_info()
- Get connection and diagnostic informationlist_team_models(team_id?)
- List all models for a teamget_model(model_id)
- Get detailed model informationlist_model_versions(model_id)
- List all versions of a modellist_deployments(team_id?)
- List all deploymentslist_preprocessors(team_id?)
- List all preprocessorsget_preprocessor(preprocessor_id)
- Get preprocessor detailsget_collection_scenarios(collection_id)
- List scenarios in a collectionget_active_team_deploy_keys_count(team_id?)
- Get count of active deploy keysmisc_get_version_info()
- Get version information
Write Tools (Restricted)
Note: Write tools require ENABLE_WRITE_TOOLS=true
in environment
activate_deployment(deployment_id)
- Activate a deploymentdeactivate_deployment(deployment_id)
- Deactivate a deploymentgenerate_deploy_key(deployment_id, description?, days_until_expiry?)
- Generate deployment keyget_deployment_payload(deployment_id)
- Get sample payload data for deploymentgpt_generate_report(model_id, version_id, ...)
- Generate GPT report
Security
Authentication
The server requires authentication via:
- Bearer tokens for MCP client connections
- API keys for Xplainable backend (from environment only)
Transport Security
- Default binding to localhost only
- TLS termination at reverse proxy recommended
- Origin/Host header validation
Rate Limiting
Per-tool and per-principal rate limits are enforced to prevent abuse.
Synchronization with xplainable-client
When the xplainable-client library is updated, use these tools to keep the MCP server synchronized:
Quick Sync Check
Comprehensive Sync Process
- Read the sync workflow guide:
SYNC_WORKFLOW.md
- Review common scenarios:
examples/sync_scenarios.md
- Run automated analysis:
python scripts/sync_workflow.py
- Implement changes following the patterns in
server.py
- Test thoroughly and update documentation
Development
Setup
Testing
Deployment
Docker
Compatibility Matrix
MCP Server Version | Xplainable Client | Backend API |
---|---|---|
0.1.x | >=1.0.0 | v1 |
Contributing
See CONTRIBUTING.md for guidelines.
Security
For security issues, please see SECURITY.md.
License
MIT License - see LICENSE for details.
This server cannot be installed
remote-capable server
The server can be hosted and run remotely because it primarily relies on remote services or has no dependency on the local environment.
Enables secure access to Xplainable AI platform capabilities for managing machine learning models, deployments, and preprocessors. Supports both read operations (listing models, deployments) and write operations (deploying models, generating reports) with proper authentication and rate limiting.
Related MCP Servers
- AsecurityFlicenseAqualityEnables AI assistants to manage Xano databases through the Model Context Protocol, allowing users to create, modify, and delete tables, edit schemas, and extract API documentation.Last updated -97
- -securityFlicense-qualityAllows AI models to query and retrieve analytics data from Plausible Analytics through the Plausible API, enabling natural language interactions with website statistics.Last updated -
- -securityAlicense-qualityEnables AI assistants to interact with OKX trading accounts through read-only access to retrieve portfolio information, trading positions, order history, and account analytics. Provides secure, local processing of trading data without storing sensitive information or enabling trade execution.Last updated -63MIT License
- -securityFlicense-qualityEnables AI agents to manage Linear issues, projects, teams, users, comments, and cycles through an optimized interface designed specifically for language models. Supports both local and remote deployment with OAuth authentication and batch operations.Last updated -216