athena-protocol

Athena Protocol MCP Server is an intelligent validator and guidance engine designed to assist AI coding agents. It acts as a senior engineer, providing systematic thinking validation, impact analysis, assumption checks, dependency mapping, and thinking optimization before code changes are made. The server supports multiple LLM providers (14 total) and is designed to operate with environment-driven configuration, ensuring no hardcoded defaults. When running locally or via MCP clients, you can point your client configuration at the Athena server and rely on the advanced file analysis capabilities (including precision-targeted analysis with analysisTargets) to reduce token usage and speed up validations. It also offers session-based validation history, health checks, and a dual-agent architecture to improve throughput and reliability across validation tasks.

Prerequisites:

• Node.js >= 18
• npm (or yarn)

1. Clone the repository or download the package sources.
2. Install dependencies and build:

npm install
npm run build

3. Prepare configuration:

• Create and edit a .env file with your provider settings (API keys, model names, etc.). See .env.example for available options.
• Ensure PROVIDER_SELECTION_PRIORITY is defined in .env (or environment) to prioritize your LLM providers.

4. Validate the configuration and run:

npm install
npm run build
npm run validate-config  # Validates your .env configuration
npm test

5. Start the server (example):

npm run build
npm start

6. Optional: If using MCP client configurations, reference the local path to the built server (dist/index.js) in your client mcpServers configuration.

Notes:

• The server expects environment-driven configuration; do not rely on hardcoded defaults.
• For NPM-based usage, you can also run via npx with environment variables as shown in the client configuration examples.

Tips and notes:

• The Athena Protocol relies entirely on environment variables for configuration; ensure PROVIDER_SELECTION_PRIORITY is set and that each provider has its API keys and model parameters configured.
• GPT-5 specific parameters are supported for advanced reasoning models; some defaults are deprecated or not used, so rely on the documented environment variables in your .env or environment.
• When using npx for MCP clients, prefer the environment-driven configuration in the client examples to ensure correct provider keys and model settings.
• If you encounter startup validation failures, check .env for missing keys or misconfigured provider blocks, and use npm run validate-config to catch issues early.