Collected sources and patterns will appear here. Add from search, explore, or the patterns library.
SystemConfig -> CompletionClient
Map diverse LLM architectures, pipeline chains, or agent runtimes to a uniform prompt-in, completion-out protocol for standardized execution.
Problem it solves
Evaluating diverse target systems (raw LLMs, prompt chains, agents) requires a shared calling interface to avoid writing customized evaluation loops for each system.
Consumes
Emits
The real projects this mechanism was found in. Attribution is the point — this is how the best teams actually do it.