SuperOptiX
Full-Stack Agent Optimization
The first universal agent optimization framework. Build once with SuperSpec, compile to any framework, optimize every layer with GEPA.
Write Once, Compile Everywhere
8 Frameworks. One SuperSpec.
Why SuperOptiX?
Agents fail in production because optimization is an afterthought. SuperOptiX makes it the foundation.
The Problem
- Agents fail unpredictably in production
- Manual prompt tweaking is endless
- No systematic optimization exists
- Frameworks ignore the core issue
The Reality
Without systematic optimization, AI agents remain research prototypes. Optimization at every layer—prompts, context, memory, tools—is what transforms agents into production-ready systems.
The Gap
90% of agent frameworks focus on orchestration. 0% focus on optimization.
SuperOptiX
- Optimization is our entire focus
- Built for production from day one
- Industry-leading GEPA optimizer
- Works with any framework or custom agents
How It Works
The Agent Building Factory
Declare once, compile anywhere, optimize everything. From spec to production in minutes.
Declare Agent
Write your agent as a SuperSpec YAML specification. Define behavior, tools, and goals once.
super agent pull developerCompile Anywhere
Generate native pipelines for any supported framework. One spec, multiple runtimes.
super agent compile developer --framework dspyRun & Evaluate
Execute your agent with built-in evaluation metrics. Know how your agent performs.
super agent run developer --goal "..."Optimize
GEPA optimizes across all layers. Enable with --optimize when ready for production.
super agent optimize developer --auto lightGEPA Optimizer
Optimization at Every Layer
SuperOptiX optimizes across prompts, context, memory, and pipelines to make agents production-worthy.
Prompt Optimization
Automatic prompt evolution and refinement using DSPy optimizers
Context Engineering
Structured context management with intelligent chunking and retrieval
Pipeline Optimization
End-to-end workflow refinement with GEPA-powered optimization loops
Model Tuning
Hyperparameter optimization and model selection for peak performance
Quick Start
$ uv tool install superoptix
$ super agent pull developer
$ super agent compile developer --framework dspy
$ super agent run developer --framework dspy --goal "Design a migration strategy"
# Optional optimization path
$ super agent compile developer --framework dspy --optimize
$ super agent optimize developer --framework dspy --auto lightOpen Source Release
What Ships in This Release
SuperOptiX is now publicly available for the community to use, inspect, and extend.
Open Source
SuperOptiX is publicly available on GitHub. Use it, inspect the code, contribute, and build on top of it.
View on GitHubMinimal Pipelines
Generated pipelines are significantly simpler and runnable by default. No boilerplate.
Explicit Optimize Path
Optimization and evaluation logic lives behind --optimize so base pipelines stay clean.
8 Framework Support
Native generation flows across DSPy, Pydantic AI, Google ADK, OpenAI, DeepAgents, CrewAI, Claude SDK, and Microsoft.
RLM Support (Experimental)
Experimental RLM support with stronger Signature + Module automation and GEPA-first workflow.
StackOne Connectors
Wire SaaS actions into agents using StackOne connectors for connector-driven automation.
Ready to Build Production-Ready Agents?
Join the community of developers building optimized agents with SuperOptiX.