Athena Protocol

🚀 Athena Protocol MCP Server: A Precursor to Context Orchestration Protocol

An intelligent MCP server that serves as an AI tech lead for coding agents. It offers expert validation, impact analysis, and strategic guidance before code changes. Similar to a senior engineer reviewing your approach, the Athena Protocol helps AI agents identify critical issues early, validate assumptions against the actual codebase, and optimize their problem - solving strategies. The outcome is higher - quality code, fewer regressions, and more thoughtful architectural decisions.

Key Feature: Precision file analysis with analysisTargets can achieve a 70 - 85% token reduction and 3 - 4× faster performance through precision - targeted code analysis. See Enhanced File Analysis for details.

🚀 Quick Start

The Athena Protocol MCP Server is an intelligent tool for AI coding agents. It provides validation and guidance before code changes, improving code quality. To get started, you need to install the necessary dependencies and configure the environment.

✨ Features

Context Orchestration Protocol: The Future of AI - Assisted Development

Imagine LLMs operating with highly refined and targeted context. This eliminates guesswork, reduces errors by 80%, and enables the delivery of code with the precision of experienced architects. It transforms how AI agents understand and enhance complex codebases.

Key Features of Athena Protocol MCP Server

• Precision File Analysis: With analysisTargets, it can achieve 70 - 85% token reduction and 3 - 4× faster performance.
• Smart Client Mode: Precision - targeted code analysis with significant token reduction.
• Environment - Driven Configuration: No hardcoded defaults, all configured through the .env file.
• Multi - Provider LLM Support: Supports 14 LLM providers with automatic fallback.
• Enhanced File Reading: Multiple modes including full, head, tail, and range.
• Concurrent File Operations: Improves performance by 3 - 4×.
• Session - Based Validation History and Memory Management: Keeps track of validation history and manages memory effectively.
• Comprehensive Configuration Validation and Health Monitoring: Ensures proper configuration and monitors system health.
• Dual - Agent Architecture: For efficient validation workflows.

📦 Installation

This module requires knowledge of Node.js and npm.

npm install
npm run build

Prerequisites

• Node.js >= 18
• npm or yarn

Configuration

The Athena Protocol uses 100% environment - driven configuration, with no hardcoded provider values or defaults. Configure everything through your .env file:

1. Copy the example configuration:

cp .env.example .env

2. Edit .env and configure your provider:

   • Set DEFAULT_LLM_PROVIDER (e.g., openai, anthropic, google).
   • Add your API key for the chosen provider.
   • Configure model and parameters (optional).

3. Validate and test:

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

See .env.example for complete configuration options and all 14 supported providers.

Critical Configuration Requirements

• PROVIDER_SELECTION_PRIORITY is REQUIRED. List your providers in priority order.
• No hardcoded fallbacks exist. All configuration must be explicit in .env.
• Fail - fast validation. Invalid configuration causes immediate startup failure.
• Complete provider config is required, including API key, model, and parameters for each provider.

Supported Providers

The Athena Protocol supports 14 LLM providers. While OpenAI is commonly used, you can configure any of the following:

Major Cloud Providers:

• OpenAI - GPT - 5 (with thinking), GPT - 4o, GPT - 4 - turbo
• Anthropic - Claude Opus 4.1, Claude Sonnet 4.5, Claude Haiku 4.5
• Google - Gemini 2.5 (Flash/Pro/Ultra)
• Azure OpenAI - Enterprise - grade GPT models
• AWS Bedrock - Claude, Llama, and more
• Google Vertex AI - Gemini with enterprise features

Specialized Providers:

• OpenRouter - Access to 400+ models
• Groq - Ultra - fast inference
• Mistral AI - Open - source models
• Perplexity - Search - augmented models
• XAI - Grok models
• Qwen - Alibaba's high - performance LLMs
• ZAI - GLM models

Local/Self - Hosted:

• Ollama - Run models locally

Quick switch example:

# Edit .env file
ANTHROPIC_API_KEY=sk - ant - your - key - here
DEFAULT_LLM_PROVIDER=anthropic

# Restart server
npm run build && npm start

Provider Switching

See the detailed provider guide for complete setup instructions.

💻 Usage Examples

MCP Client Configuration

For detailed, tested MCP client configurations, see CLIENT_MCP_CONFIGURATION_EXAMPLES.md

For Local Installation (with .env file)

Local installation with a .env file remains fully functional and unchanged. Simply clone the repository and run:

npm install
npm run build

Then configure your MCP client to point to the local installation:

{
  "mcpServers": {
    "athena - protocol": {
      "command": "node",
      "args": ["/absolute/path/to/athena - protocol/dist/index.js"],
      "type": "stdio",
      "timeout": 300
    }
  }
}

For NPM Installation (with MCP environment variables - RECOMMENDED)

For npm/npx usage, configure your MCP client with environment variables. Only the configurations in CLIENT_MCP_CONFIGURATION_EXAMPLES.md are tested and guaranteed to work.

Example for GPT - 5:

{
  "mcpServers": {
    "athena - protocol": {
      "command": "npx",
      "args": ["@n0zer0d4y/athena - protocol"],
      "env": {
        "DEFAULT_LLM_PROVIDER": "openai",
        "OPENAI_API_KEY": "your - openai - api - key - here",
        "OPENAI_MODEL_DEFAULT": "gpt - 5",
        "OPENAI_MAX_COMPLETION_TOKENS_DEFAULT": "8192",
        "OPENAI_VERBOSITY_DEFAULT": "medium",
        "OPENAI_REASONING_EFFORT_DEFAULT": "high",
        "LLM_TEMPERATURE_DEFAULT": "0.7",
        "LLM_MAX_TOKENS_DEFAULT": "2000",
        "LLM_TIMEOUT_DEFAULT": "30000"
      },
      "type": "stdio",
      "timeout": 300
    }
  }
}

See CLIENT_MCP_CONFIGURATION_EXAMPLES.md for complete working configurations.

Configuration Notes:

• NPM Installation: Use npx @n0zer0d4y/athena - protocol with the env field for the easiest setup.
• Local Installation: Local .env file execution remains fully functional and unchanged.
• Environment Priority: MCP env variables take precedence over .env file variables.
• GPT - 5 Support: Includes specific parameters for GPT - 5 models.
• Timeout Configuration: The default timeout of 300 seconds (5 minutes) is set for reasoning models like GPT - 5. For faster LLMs (GPT - 4, Claude, Gemini), you can reduce this to 60 - 120 seconds.
• GPT - 5 Parameter Notes: The parameters LLM_TEMPERATURE_DEFAULT, LLM_MAX_TOKENS_DEFAULT, and LLM_TIMEOUT_DEFAULT are currently required for GPT - 5 models but are not used by the model itself. This is a temporary limitation that will be addressed in a future refactoring.
• Security: Never commit API keys to version control. Use MCP client environment variables instead.

Future Refactoring Plans

GPT - 5 Parameter Optimization

Current Issue: GPT - 5 models currently require the standard LLM parameters (LLM_TEMPERATURE_DEFAULT, LLM_MAX_TOKENS_DEFAULT, LLM_TIMEOUT_DEFAULT) even though these parameters are not used by the model.

Planned Solution:

1. Modify getTemperature() function to return undefined for GPT - 5+ models instead of a hardcoded default.
2. Update AI provider interfaces to handle undefined temperature values.
3. Implement conditional parameter validation that skips standard parameters for GPT - 5+ models.
4. Update OpenAI provider to omit unused parameters when communicating with GPT - 5 API.

Benefits:

• Cleaner configuration for GPT - 5 users.
• More accurate representation of model capabilities.
• Better adherence to OpenAI's GPT - 5 API specification.

Timeline: Target implementation in v0.3.0

Server Modes

MCP Server Mode (for production use)

npm start                    # Start MCP server for client integration (requires .env or MCP env)
npm run dev                  # Development mode with auto - restart
npx @n0zer0d4y/athena - protocol  # Run published version via npx (requires MCP env)

Standalone Mode (for testing)

npm run start:standalone     # Test server without MCP client
npm run dev:standalone       # Development standalone mode

Configuration Tools

# Validate your complete configuration
npm run validate - config

# Or use the comprehensive MCP validation tool
node dist/index.js
# Then call: validate_configuration_comprehensive

📚 Documentation

API

The Athena Protocol MCP Server provides the following tools for thinking validation and analysis:

thinking_validation

Validate the primary agent's thinking process with focused, essential information.

Required Parameters:

• thinking (string): Brief explanation of the approach and reasoning.
• proposedChange (object): Details of the proposed change.
  • description (string, required): What will be changed.
  • code (string, optional): The actual code change.
  • files (array, optional): Files that will be affected.
• context (object): Context for the validation.
  • problem (string, required): Brief problem description.
  • techStack (string, required): Technology stack (react|node|python etc).
  • constraints (array, optional): Key constraints.
• urgency (string): Urgency level (low, medium, or high).
• projectContext (object): Project context for file analysis.
  • projectRoot (string, required): Absolute path to project root.
  • workingDirectory (string, optional): Current working directory.
  • analysisTargets (array, REQUIRED): Specific code sections with targeted reading.
    • file (string, required): File path (relative or absolute).
    • mode (string, optional): Read mode - full, head, tail, or range.
    • lines (number, optional): Number of lines (for head/tail modes).
    • startLine (number, optional): Start line number (for range mode, 1 - indexed).
    • endLine (number, optional): End line number (for range mode, 1 - indexed).
    • priority (string, optional): Analysis priority - critical, important, or supplementary.
• projectBackground (string): Brief project description to prevent hallucination.

Optional Parameters:

• sessionId (string): Session ID for context persistence.
• provider (string): LLM provider override (openai, anthropic, google, etc).

Output: Returns validation results with confidence score, critical issues, recommendations, and test cases.

impact_analysis

Quickly identify key impacts of proposed changes.

Required Parameters:

• change (object): Details of the change.
  • description (string, required): What is being changed.
  • code (string, optional): The code change.
  • files (array, optional): Affected files.
• projectContext (object): Project context (same structure as thinking_validation).
  • projectRoot (string, required).
  • analysisTargets (array, REQUIRED): Files to analyze with read modes.
  • workingDirectory (optional).
• projectBackground (string): Brief project description.

Optional Parameters:

• systemContext (object): System architecture context.
  • architecture (string): Brief architecture description.
  • keyDependencies (array): Key system dependencies.
• sessionId (string): Session ID for context persistence.
• provider (string): LLM provider override.

Output: Returns overall risk assessment, affected areas, cascading risks, and quick tests to run.

assumption_checker

Rapidly validate key assumptions without over - analysis.

Required Parameters:

• assumptions (array): List of assumption strings to validate.
• context (object): Validation context.
  • component (string, required): Component name.
  • environment (string, required): Environment (production, development, staging, testing).
• projectContext (object): Project context (same structure as thinking_validation).
  • projectRoot (string, required).
  • analysisTargets (array, REQUIRED): Files to analyze with read modes.
• projectBackground (string): Brief project description.

Optional Parameters:

• sessionId (string): Session ID for context persistence.
• provider (string): LLM provider override.

Output: Returns valid assumptions, risky assumptions with mitigations, and quick verification steps.

dependency_mapper

Identify critical dependencies efficiently.

Required Parameters:

• change (object): Details of the change.
  • description (string, required): Brief change description.
  • files (array, optional): Files being modified.
  • components (array, optional): Components being changed.
• projectContext (object): Project context (same structure as thinking_validation).
  • projectRoot (string, required).
  • analysisTargets (array, REQUIRED): Files to analyze with read modes.
• projectBackground (string): Brief project description.

Optional Parameters:

• sessionId (string): Session ID for context persistence.
• provider (string): LLM provider override.

Output: Returns critical and secondary dependencies, with impact analysis and test focus areas.

thinking_optimizer

Optimize thinking approach based on problem type.

Required Parameters:

• problemType (string): Type of problem (bug_fix, feature_impl, or refactor).
• complexity (string): Complexity level (simple, moderate, or complex).
• timeConstraint (string): Time constraint (tight, moderate, or flexible).
• currentApproach (string): Brief description of current thinking.
• projectContext (object): Project context (same structure as thinking_validation).
  • projectRoot (string, required).
  • analysisTargets (array, REQUIRED): Files to analyze with read modes.
• projectBackground (string): Brief project description.

Optional Parameters:

• sessionId (string): Session ID for context persistence.
• provider (string): LLM provider override.

Output: Returns a comprehensive optimization strategy including:

• optimizedStrategy: Recommended approach, tools to use, time allocation breakdown, success probability, and key focus areas.
• tacticalPlan: Detailed implementation guidance with problem classification, grep search strategies, key findings hypotheses, decision points, step - by - step implementation plan, testing strategy, risk mitigation, progress checkpoints, and value/effort assessment.
• metadata: Provider used and file analysis metrics.

athena_health_check

Check the health status and configuration of the Athena Protocol server.

Parameters: None

Output: Returns default provider, list of active providers with valid API keys, configuration status, and system health information.

session_management

Manage thinking validation sessions for context persistence and progress tracking.

Required Parameters:

• action (string): Session action - create, get, update, list, or delete.

Optional Parameters:

• sessionId (string): Session ID (required for get, update, delete actions).
• tags (array): Tags to categorize the session.
• title (string): Session title/description (for create/update).

Output: Returns session information or list of sessions depending on the action.

Enhanced File Analysis (NEW)

All tools now support Smart Client Mode with analysisTargets for precision targeting:

Benefits:

• 70 - 85% token reduction by reading only relevant code sections.
• 3 - 4× faster with concurrent file reading.
• Mode - based reading: full, head (first N lines), tail (last N lines), range (lines X - Y).
• Priority processing: critical → important → supplementary.

Example:

{
  "projectContext": {
    "projectRoot": "/path/to/project",
    "analysisTargets": [
      {
        "file": "src/auth.ts",
        "mode": "range",
        "startLine": 45,
        "endLine": 78,
        "priority": "critical"
      },
      {
        "file": "src/config.ts",
        "mode": "head",
        "lines": 20,
        "priority": "supplementary"
      }
    ]
  }
}

Note: All tools require analysisTargets for file analysis. Provide at least one file with appropriate read mode (full, head, tail, or range).

🔧 Technical Details

Memory System Status

The persistent memory system (thinking - memory.json) is currently under review and pending refactoring. While functional, it:

• Creates a memory file in the project root directory.
• Persists validation history across sessions.
• May require manual cleanup during testing/development.

Planned improvements:

• Move storage to a .gitignore'd directory (e.g., athena - memory/).
• Add automatic cleanup mechanisms.
• Enhance session management.
• Improve file path handling.

For production use, consider this feature as experimental until the refactor is complete.

Configuration Methods

Athena Protocol supports two configuration methods with clear priority ordering:

1. MCP Client Environment Variables (highest priority - recommended for npm installations).
2. Local .env File (fallback - for local development).
3. System Environment Variables (lowest priority).

For npm - published usage, configure all settings directly in your MCP client's env field. For local development, continue using .env files.

Provider Testing Status

While Athena Protocol supports 14 LLM providers, only the following have been thoroughly tested:

• OpenAI
• Google
• ZAI
• Mistral
• OpenRouter
• Groq

Other providers (Anthropic, Qwen, XAI, Perplexity, Ollama, Azure, Bedrock, Vertex) are configured and should work, but have not been extensively tested. If you encounter issues with any provider, please [open an issue](https://github.com/n0zer0d4y/athena - protocol/issues) with:

• Provider name and model.
• Error messages or unexpected behavior.
• Your MCP configuration or .env configuration (redact API keys).

🤝 Contributing

This server is designed specifically for LLM coding agents. Contributions should focus on:

• Adding new LLM providers.
• Improving validation effectiveness.
• Enhancing context awareness.
• Expanding validation coverage.
• Optimizing memory management.
• Adding new validation strategies.

📄 License

MIT License - see LICENSE file for details.