What problem does it solve?
Enable KAgentic users to integrate new language model providers into the framework so agents can call external or local LLM APIs without ad-hoc hacks. The skill removes uncertainty about authentication, request/response formats, retry behavior, and testing when adding a new provider to a Kotlin JVM agent.
Core Features & Use Cases
- Implement the LLMProvider interface with a single generate(input: String) method and a configurable model selection pattern.
- Provide authentication patterns (bearer token, custom headers, URL key, or no auth for local), JSON request building, and robust response parsing for varied provider schemas.
- Add retry logic using exponential backoff, unit tests, and documentation entries so the provider is production-ready and maintainable.
- Use Case: Integrate Mistral, Together AI, Perplexity, Gemini, Claude, or a local Ollama instance to enable chat and completion functionality in KAgentic-based agents.
Quick Start
Implement a new Kotlin class that implements LLMProvider, wire an HTTP client with the provider's required authentication, add retryWithBackoff for resilience, write basic unit tests, and update README and CHANGELOG.