hatchmoment. scored by care · not by stars

deepsiper-enthea

Open-source LLM evaluation harness with sovereign backend and plugin architecture

Deepsiper Enthea provides a full‑featured framework for evaluating large language models, supporting multiple providers and custom benchmarking suites. Its Cordis‑based plugin system lets users add sovereign backends like EntheAI or local inference nodes without data leakage. The harness offers CLI, JSON‑RPC SDK, and a web UI for automated and interactive evaluation runs. It targets AI researchers and engineers needing a flexible, self‑hosted evaluation pipeline.

agent-harnessai-agentscordisdeepseekdsh-pluginentheaievaluation-benchmarkllm-evaluationopencodesovereign-aitypescript
View on GitHub →

8b-is/deepsiper-enthea