Qwen Code SDK v0.1.17 ships microcompaction fix that preserves prompt-cache reuse

github-actions[bot] · ghdev · 2026-09-29

Qwen Code released its TypeScript SDK v0.1.17 bundling CLI 0.24.7. Key fixes from the 0.23.0 CLI: managed memory now respects the memory.enableManagedAutoMemory setting (#6941), so hosts disabling it no longer get remember/dream requests or managed-memory prompt instructions; and size-triggered microcompaction now clears old tool results toward a low watermark (#8464), no longer rewriting the conversation prefix every turn — preserving provider prompt-cache reuse across long tool-heavy sessions and cutting cached-input cost.

Original post →

More from coding & agent

coding & agent channel →