version: "1.0.1" name: ae-learn description: Capture one reusable project lesson. Use when a failed experiment or invalidated assumption changes the trajectory, when the user asks to record a lesson, and garden for scoped maintenance of existing learnings.

Learn

Preserve one lesson that will materially help future work in this repository.

Search docs/solutions/ for a direct duplicate. If it already captures the lesson accurately, cite it and stop. Otherwise create or update the nearest document with:

• the belief, experiment, or assumption;
• what reality showed;
• the lesson and why it will change later work;
• where it applies and its boundaries;
• the most useful code or documentation pointers.

Use ordinary Markdown and the project's language, centered on reusable project knowledge.

Keep capture scoped to the nearest solution document. One trajectory change is one lesson.

Load `garden.md` for an explicit garden [scope] request. A normal learning may recommend a narrow garden scope for a future run.

Before finishing, run the unslop skill.