Skip to main content

​
How Users run an Agent

  1. Navigate to the Agent Page
    • Go to the Agent listings
    • Select your desired Agent
  2. Click “Run” Button
    • Located at the top of the Agent details page
    • Initiates the Agent execution process
  3. Follow On-Screen Instructions
    • Read carefully through each step
    • Input required parameters when prompted
    • Verify entered information before proceeding

​
Expectations from Creators

​
Required Documentation

  • Agent-specific documentation should include comprehensive details that enable users to effectively implement and utilize the agent
  • Clear input/output specifications detailing all possible parameters and their effects
  • Detailed step-by-step execution instructions for various use cases and scenarios
  • In-depth explanation of the underlying agent architecture and workflow patterns
  • Comprehensive list of supported LLM models with their specific configuration requirements
  • Detailed integration instructions for various platforms (Slack, Gmail, etc.) with example implementations
  • Specify compatible models and versions. Include LLM configuration being used and performance characteristics
  • Add complete details on Integrations (Slack, GoogleSheet etc.) used in the agent. Clearly mention the usage & scope of their API usage.

​
Input/Output Guidelines

  • Input Parameters should be thoroughly documented:
    • Clear distinction between required parameters that must be provided and optional ones that enhance functionality
    • Detailed specifications of parameter formats, including data types, ranges, and validation rules
    • Real-world examples demonstrating input usage in common business scenarios
  • Expected Outputs need clear documentation:
    • Detailed specifications of output formats including all possible response structures
    • Multiple sample responses covering both successful and error scenarios
    • Comprehensive error handling strategies and fallback mechanisms for edge cases

​
Best Practices

  • Detailed usage recommendations based on real-world implementations
  • Common pitfalls and their solutions, backed by practical examples
  • Comprehensive troubleshooting guide with diagnostic procedures and resolution steps

​
AfterMath

The platform implements a multi-dimensional evaluation framework that assesses agents on accuracy, efficiency, adaptability, and user satisfaction. Each metric is calculated using standardized tests appropriate to the agent’s domain and complexity level, ensuring fair comparisons across different implementations. Evaluation results are updated monthly and include confidence intervals to indicate performance stability across different usage scenarios.
Detailed comparison matrices highlight the strengths and limitations of agents addressing similar business problems. The analysis includes response time distributions, token efficiency metrics, and capability coverage maps that visualize functional overlaps and unique features. Specialized comparison views enable users to prioritize factors most relevant to their specific use cases and business constraints.
Verified user reviews include usage duration, implementation context, and specific outcomes achieved with supporting evidence. The platform distinguishes between reviews from casual users, power users, and enterprise implementations to provide context-appropriate feedback. A reputation system rewards constructive, detailed reviews while filtering out low-quality or potentially biased feedback.
Industry-specific benchmark suites test agent performance against real-world scenarios derived from actual business challenges. Regular benchmark updates reflect evolving industry requirements and new technological capabilities. Customizable benchmark reports allow organizations to evaluate agent performance specifically for their unique business environment and constraints.