Skip to content

Harden memu retrieval and topic scoring fallbacks#33

Open
elinawang626-design wants to merge 1 commit intoNevaMind-AI:mainfrom
elinawang626-design:codex/memory-fallback-fixes-clean
Open

Harden memu retrieval and topic scoring fallbacks#33
elinawang626-design wants to merge 1 commit intoNevaMind-AI:mainfrom
elinawang626-design:codex/memory-fallback-fixes-clean

Conversation

@elinawang626-design
Copy link
Copy Markdown

Summary

This PR tightens two fallback paths that currently fail silently or behave inconsistently on main.

Changes

  • route proactive memu_memory through the shared memu executor so proactive mode and normal chat use the same retrieval path
  • require complete memU configuration before attempting remote retrieval
  • return explicit errors for non-OK memU retrieval responses instead of treating them as successful results
  • fall back to heuristic temporary-topic scoring when no Claude API key is configured
  • add regression tests for memU retrieval error handling and heuristic temporary-topic transitions

Verification

  • npx vitest run src/main/tools/__tests__/memu.executor.test.ts src/main/services/agent/context/layered/__tests__/temporary-topic.test.ts

Notes

  • I also ran broader repo checks locally and saw existing unrelated failures in the current upstream codebase (for example in secure-storage and several typecheck errors), so this PR keeps verification focused on the changed paths.

Sign up for free to join this conversation on GitHub. Already have an account? Sign in to comment

Labels

None yet

Projects

None yet

Development

Successfully merging this pull request may close these issues.

1 participant