DeepYardDeepYard
B

BackendForge

Benchmark framework for evaluating AI agents on end-to-end backend code generation

Open SourceFree

About

BackendForge is a research benchmark designed to evaluate LLM agents on their ability to generate complete, deployable backend services. Unlike traditional code completion benchmarks, it tests agents on creating full software artifacts with correct API behavior and deployment configurations. Includes controlled scenarios with predefined API contracts, making it valuable for researchers assessing agentic coding capabilities beyond single-file generation.

Details

Type
Integrations
Language

Tags

coding-agentevaluationopen-sourceautonomousframework