model-service

What problem does it solve?

Use this skill whenever the task is to call LLM APIs through this repository by writing or adjusting model-service YAML config and running the existing pipeline. This skill should trigger for requests like "call model/api", "write yaml config", "run model services", "test provider/model route", "send prompt/files to llm", or OpenRouter/Gemini/Kimi/Aliyun/Seed invocation. Strong rule: in this skill, modify YAML config only and do not modify Python scripts.

Core Features & Use Cases

• YAML-based configuration for model-service pipelines to drive API calls across providers (openrouter, gemini, kimi, aliyun, seed)
• Environment-driven credentials: URLs and keys sourced from env.txt and referenced via $VAR or ${VAR}
• Guardrails: if a provider's credentials resolve to empty, it is treated as unsupported in the current environment
• Safe boundaries: editing YAML config only; Python scripts are not modified

Quick Start

Edit the YAML config at .github/scripts/configs/experiments/model_services_template.yaml and run the provided Python snippet to load the config and execute the pipeline.