mirror of
https://github.com/lobehub/lobehub.git
synced 2026-03-26 13:19:34 +07:00
* ♻️ refactor: add eval-only server context compression * ♻️ refactor: align eval compression with runtime step flow * ♻️ refactor: trim redundant call_llm diff * ✨ add mid-run context compression step * 📝 document post compression helper * 🐛 revert unnecessary agent runtime service diff * ♻️ refactor: clean up context compression follow-up logic * ♻️ refactor: move compression gate before call llm * ♻️ refactor: make call llm compression gate explicit * ♻️ refactor: restore agent-side compression checks * ♻️ refactor: rename agent llm continuation helper * ♻️ refactor: inline agent compression helper * ♻️ refactor: preserve trailing user message during compression * 📝 docs: clarify toLLMCall refactor direction * ✅ test: add coverage for context compression flow * ⏪ reset: unstash