continuous-learning

Capture reusable knowledge by identifying significant discoveries, condensing them into durable documentation, and registering entries in a learnings registry with type, source, summary, location, and confidence to prevent knowledge loss.

Prevent knowledge loss by classifying learnings as skill, agent behavior, workflow, or reference types, then creating or updating corresponding files under .opencode directories and appending them to the learnings registry.

Classify learnings by type to determine storage: generate skill entries under .opencode/skills, update agent instructions under .opencode/agent, and record long-term reference docs under .opencode/docs and .maestro.

A GitHub repository provides the structured file system needed to store .opencode skills, agent instructions, and the learnings registry, ensuring workflow improvements are tracked and discoverable for future development phases.

Create a new skill entry under .opencode/skills when capturing a distinct reusable capability, and update agent behavior under .opencode/agent when adjusting existing processing logic or workflow rules.

Yes, a learning registry manages agent behavior adjustments and workflow improvements by recording each learning's type, source, summary, location, and confidence level for consistent discoverability and future reference.