model-registry-maintainer

The Model Registry Maintainer is a specialized utility for developers managing the MassGen ecosystem. It streamlines the lifecycle management of AI models by ensuring that the internal registry remains synchronized with provider releases, API updates, and cost-efficiency requirements. This skill is intended for technical contributors who need to incorporate new LLM providers, adjust model parameters, or calibrate token management logic.

• Updates massgen/backend/capabilities.py to manage model availability, supported capabilities like web search or code execution, and model release dates.

• Manages massgen/token_manager/token_manager.py to ensure accurate cost estimation, token counting, and context window limits for models not found in the LiteLLM database.

• Provides a structured workflow for gathering critical model metadata including input/output pricing per 1K tokens and exact API identifiers.

• Validates changes through integrated test suites to ensure backend integrity before deployment.

• Automates documentation generation to keep backend tables and reference guides current for end users.

• Always prioritize the LiteLLM database for pricing resolution before falling back to hardcoded PROVIDER_PRICING entries.

• Ensure exact case-sensitive model names are used as defined by provider API documentation.

• Consult official news sources from OpenAI, Anthropic, Google, and xAI when updating release dates and performance specs.

• Execute verification tests using the pytest framework to confirm configuration changes do not break system functionality.

• Use the provided scripts to regenerate documentation to ensure parity between the code and the developer portal.