A Go implementation of the DSPy framework for building reliable LLM-powered applications — featuring self-improving agents, state-of-the-art optimizers, and Go-native performance.
go get github.com/XiaoConstantine/dspy-go
Build agents with ReAct reasoning, memory management, and orchestration. ACE Framework enables agents that learn from execution—recording trajectories, extracting patterns, and improving over time. Learn more
GEPA uses evolutionary algorithms with Pareto optimization. MIPRO brings TPE-based systematic search. SIMBA adds introspective learning. 5 advanced optimizers that outperform manual prompt engineering.
Smart Tool Registry with Bayesian inference for optimal tool selection. Tool chaining with dependency resolution. MCP integration for external services. Build sophisticated agent workflows.
Built for Go from the ground up—not a port. Parallel module execution, concurrent batch processing, efficient streaming. Production-ready with retry logic and monitoring.
Start instantly with one-line zero-config setup. Try all optimizers without writing code using the CLI tool. Test with GSM8K, HotPotQA, and custom datasets in seconds.
Seamlessly switch between OpenAI, Anthropic, Google, Ollama, and more. Unified interface with provider-specific optimizations. Local and cloud models with the same API.
Install with a single command and build your first LLM application in minutes.
go get github.com/XiaoConstantine/dspy-go