continuous-learning

To capture domain knowledge and promote confirmed patterns, agents store observations in structured directories like knowledge/domain/, validate hypotheses, and promote confirmed patterns into rules.md. This iterative learning cycle converts insights into fleet-wide governance.

The continuous learning cycle workflow involves capturing observations into knowledge directories, validating hypotheses listed in hypotheses.md, and promoting confirmed patterns into reusable rules stored in rules.md for future guidance.

Yes, fleet-wide knowledge promotion requires structured knowledge directories. You must set up knowledge/domain/ folders containing knowledge.md, hypotheses.md, and rules.md to store insights and guide the promotion workflow.

Yes, single-developer knowledge captures stay local initially. As recurring patterns mature and validate through the learning cycle, they are promoted from local insights into fleet-wide reusable rules across the multi-agent environment.

The best way to organize multi-agent knowledge management is using structured directories under knowledge/INDEX.md and domain subfolders. This stores insights across knowledge/, docs/learning/, and docs/research/ domains to guide future work.