Skip to content

fix(init): include server,seekdb extras and improve LLM auto-detection#1016

Merged
wayyoungboy merged 1 commit into
oceanbase:mainfrom
lightzt99:fix/init-package-seekdb
Jun 11, 2026
Merged

fix(init): include server,seekdb extras and improve LLM auto-detection#1016
wayyoungboy merged 1 commit into
oceanbase:mainfrom
lightzt99:fix/init-package-seekdb

Conversation

@lightzt99

@lightzt99 lightzt99 commented Jun 11, 2026

Copy link
Copy Markdown
Collaborator

Summary

  • init.sh: Default POWERMEM_INIT_PACKAGE to powermem[server,seekdb] so pyseekdb is installed on fresh plugin init, fixing ModuleNotFoundError: No module named 'pyseekdb' at server startup
  • init.sh: Detect LLM provider from available API key rather than model name prefix — prevents provider/key mismatch when settings.json has e.g. model: "qwen/..." but the only available key is ANTHROPIC_API_KEY; key priority order: ANTHROPIC_AUTH_TOKENANTHROPIC_API_KEYOPENAI_API_KEYDEEPSEEK_API_KEYDASHSCOPE_API_KEY
  • init.sh: Add validate_llm_config step after .env creation — makes a minimal test API call and fails early with a clear error if provider/model/key are incompatible, instead of silently starting a broken server
  • SETUP.md: Fix non-CN model preload — add pip install huggingface_hub before snapshot_download call (mirrors the CN path which installs modelscope first)
  • SETUP.md: Fix dashboard build ordering — move the background build prompt to immediately after model download is started in background, so dashboard build runs in parallel with model download and server startup

@lightzt99 lightzt99 force-pushed the fix/init-package-seekdb branch from 4077ba1 to a6a33c9 Compare June 11, 2026 13:35
@lightzt99 lightzt99 changed the title fix(init): include server,seekdb extras in default package install fix(init): include server,seekdb extras and improve LLM auto-detection Jun 11, 2026
- Default POWERMEM_INIT_PACKAGE to powermem[server,seekdb] so pyseekdb
  is installed on fresh plugin init (fixes ModuleNotFoundError at startup)
- Detect LLM provider from available API key rather than model name prefix,
  preventing provider/key mismatch when settings.json model field uses a
  different provider prefix (e.g. qwen/...) than the available key
- Add validate_llm_config step after .env creation: makes a test API call
  and fails early with a clear error if the model is incompatible, instead
  of silently starting a broken server
- Fix SETUP.md non-CN model preload: add pip install huggingface_hub before
  snapshot_download call
- Fix SETUP.md dashboard ask ordering: move background build prompt to
  immediately after model download is started, so it runs in parallel
@lightzt99 lightzt99 force-pushed the fix/init-package-seekdb branch from a6a33c9 to 10a96eb Compare June 11, 2026 13:42
@wayyoungboy

Copy link
Copy Markdown
Member

LGTM

@wayyoungboy wayyoungboy merged commit 778e8b1 into oceanbase:main Jun 11, 2026
14 checks passed
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.

2 participants