design-compounder

Cross-sprint design learnings are consolidated by reading decisive sprint files such as design-qa.md and design-handoff.md, extracting durable insights, and writing them to session-log.md and design.md references.

Artifact-backed knowledge compounding extracts concrete lessons and provenance from sprint outcome files, ensuring durable design insights survive across future cycles without rehashing past conversations.

Provide the design sprint's compound_pending_feature_id to trigger extraction; the worker reads sprint evidence files, produces durable learnings, updates session logs and references, then clears the queued feature in tracked-work.json.

Yes, extracting cross-sprint design insights requires reading AGENTS.md, .agents-stack/tracked-work.json, and sprint decisive evidence files including design-qa.md, design-handoff.md, and design.md criteria to ensure accurate consolidation.

The best way to preserve design decisions is patching .agents-stack/reference/design.md and related scoped references with durable insights extracted from sprint outcomes, ensuring design criteria survive across future cycles.

No, design compounding applies when sprint results are reconciled and queued in compound_pending_feature_ids, requiring a valid feature ID to read decisive evidence and complete the deterministic learning handoff.