Examples

The examples/ directory ships larger, real-world AIMU programs, bigger than a notebook cell, organized by theme. Each subdirectory has its own README with run commands.

Example	What it shows	Guide
text-refinement/	A generate → judge → refine loop over text, implemented four ways: a code loop, an Agent, the EvaluatorOptimizer workflow, and simulated annealing. GPU-free, Ollama-only.	Iterative text refinement
image-refinement/	The same loop over images (diffusion + a vision evaluator), in five variants including image-to-image. Needs aimu[hf] and a GPU.	Iterative image refinement
news-summarizer/	One task (summarize recent AI news) solved four ways via --method: Agent, Chain, Parallel, and OrchestratorAgent.	Build an orchestrator
skills/	Demo SKILL.md skills (haiku-poet, unit-converter) for SkillAgent discovery, exposed as aimu.paths.skills.	Use skills

The text-refinement and image-refinement families are deliberately the same task in two modalities. Read them side by side: the difference between who drives the loop (code vs. agent vs. workflow) and which search strategy wins (greedy vs. hill-climbing vs. annealing) is independent of whether the artifact is a string or an image, a concrete take on agents vs. workflows.