Review completed work, analyze mistakes, identify repeatable improvements, and turn validated lessons into memory or reusable skills. Use after substantial tasks, failures, or review-heavy work.
Use this skill after meaningful work, not after trivial replies.
Skip it for one-off small answers with no reusable pattern.
Use the current Horbot memory layout, not ad-hoc logs:
L2/MEMORY.mdL1/HISTORY.mdL1/REFLECTION.mdIf the completed work forms a repeatable workflow, Horbot may also distill it into a user skill under the current agent's skills directory.
Use the templates in templates/ when helpful:
templates/code-review.mdtemplates/capability-assessment.mdtemplates/learning-plan.mdFocus on concrete evidence from the work that just happened.
MEMORY.mdHISTORY.mdREFLECTION.mdOnly persist learnings that are:
Do not preserve vague self-praise, unverified guesses, or secrets.
Horbot can run a background skill review after successful tool-backed work.
That mechanism works best when:
When the workflow is reusable, the system may: