feat(memory): add usage-aware auto search and backend-specific embeddings

[jinliyl]

Summary

Adds usage accounting for synthetic auto memory search messages, simplifies memory query generation around the latest user intent, and aligns ReMe Light embedding configuration across backend mapping, console UI, doctor checks, and documentation.

Updates

Auto memory search usage

• Adds estimated token usage to the synthetic memory_search assistant message created for auto memory search.
• Reads running.light_context_config.token_count_estimate_divisor and falls back to the shared default divisor when config loading fails.
• Stores the estimate in both AssistantMsg.usage and metadata.auto_memory_search_usage, marking it as estimated with input/output token details.
• Keeps the synthetic memory-search block ids in metadata so scroll/history persistence can continue excluding live-only auto-search context.

Memory query generation

• Moves _build_query into BaseMemoryManager so ADBPG and ReMe Light share one query-building path.
• Simplifies auto-search query selection to the latest non-empty user message instead of combining truncated user and assistant text.
• Removes duplicated MAX_QUERY_CHARS logic from the ADBPG and ReMe Light memory managers.

Embedding backend configuration

• Adds backend-specific embedding enablement rules aligned with AgentScope credential requirements:
  • OpenAI-compatible backends (openai, dashscope, dashscope_multimodal) require model_name and api_key, with optional base_url.
  • Gemini requires model_name and api_key, without passing base_url.
  • Ollama requires model_name and maps base_url to optional host.
• Builds the ReMe embedding credential payload according to the selected backend instead of always sending api_key and base_url.
• Keeps keyword and wikilink indexes available when vector embedding requirements are not met.
• Updates the bundled reme-ai dependency from 0.4.0.5 to 0.4.0.7.

Console and localization

• Adds an embedding backend selector in the ReMe Light memory config card.
• Shows or hides API key and base URL/host fields based on the selected backend.
• Updates embedding activation hints, raw-log search wording, rebuild-index guidance, and auto-memory interval recommendations.
• Refreshes related locale strings across supported console languages, including adding the missing Vietnamese memory configuration strings.

Doctor checks and documentation

• Updates doctor memory embedding notes so Ollama does not require an API key while remote embedding backends still require provider credentials.
• Documents supported embedding backends, enablement conditions, and credential mappings in both English and Chinese config and memory docs.
• Clarifies that vector search is disabled when backend-specific requirements are not met while non-vector indexes remain available.

Tests

• Adds coverage for backend-specific ReMe embedding enablement and credential mapping.
• Adds coverage for the shared latest-user-message query builder.
• Updates auto memory search tests to validate estimated usage metadata and the instance-based builder path.
• Updates scroll manager tests to continue verifying that synthetic auto memory search messages are not persisted to history.

[github-actions]

Welcome to QwenPaw! 🐾

Hi @jinliyl, this is your 57th Pull Request.

📋 About PR Template

To help maintainers review your PR faster, please make sure to include:

• ✅ Description - What this PR does and why
• ✅ Type of Change - Bug fix / Feature / Breaking change / Documentation / Refactoring
• ✅ Component(s) Affected - Core / Console / Channels / Skills / CLI / Documentation / Tests / CI/CD / Scripts
• ✅ Checklist:
  • Run and pass pre-commit run --all-files
  • Run and pass relevant tests (pytest or as applicable)
  • Update documentation if needed
• ✅ Testing - How to test these changes
• ✅ Local Verification Evidence:

  pre-commit run --all-files
  # paste summary result
  
  pytest
  # paste summary result

Complete PR information helps speed up the review process. You can edit the PR description to add these details.

🙌 Join Developer Community

Thanks so much for your contribution! We'd love to invite you to join the official QwenPaw developer group! You can find the Discord and DingTalk group links under the "Developer Community" section on our docs page:
https://qwenpaw.agentscope.io/docs/community

We truly appreciate your enthusiasm—and look forward to your future contributions! 😊

We'll review your PR soon.

---

Tip

⭐ If you find QwenPaw useful, please give us a Star!

Staying ahead

Star QwenPaw on GitHub and be instantly notified of new releases.

Your star helps more developers discover this project! 🐾