Grant customer.admin and customer.user roles access to llmage v1 API:
- /v1/chat/completions
- /v1/video/generations
- /v1/image/generations
- /v1/models
- /v1/tasks
Updated both load_path.py and setup_llmage_perms.sh
DBPools() was called without config.databases, causing NoneType error.
Load config from sage_root and pass config.databases like the other
migration script does.
Replace 6-table JOIN with 3-step approach:
1. get_llmage_llm() for base info (llm + llm_api_map + llmcatelog)
2. Cached uapi lookup (ioid, stream, callbackurl)
3. Cached uapiio lookup (input_fields)
Benefits:
- Code reuse: eliminates duplicate SQL
- Performance: uapi/uapiio cached with 5min TTL
- Maintainability: separate concerns for model info vs API config
- Adds invalidate_uapi_cache() for config changes
- Remove bgcolor/color/border hardcoded dark theme values from index.ui
- Use css:'card' class instead of inline bgcolor for navigation cards
- Remove conflicting bgcolor:#def0f0 from show_llms/show_llms_by_providers/show_same_catelog_llm
- Let system theme (bricks.css/shell_theme.css) handle styling
- Migrate all llmage permission entries from sage/load_path.py
- Include new data_filter API endpoints (llm_list/create/update/delete, get_organizations, get_upapps)
- Include all existing v1, api, CRUD directory, and page endpoints
- Follow product_management/scripts/load_path.py pattern
- Add data_filter with 4 searchable fields (name LIKE, model LIKE, providerid, upappid)
- Add filter_labels for search form display
- Create llm_list.dspy with DBFilter support and LIKE wildcard handling
- Create llm_create.dspy, llm_update.dspy, llm_delete.dspy
- Create get_organizations.dspy and get_upapps.dspy for dropdown options
- Add browserfields alters for providerid and upappid dropdowns
- Add editable URLs for DataViewer CRUD operations
Removed the 'if llm.ownerid == userorgid: return True' shortcut from
checkCustomerBalance() in llmage/accounting.py. All requests now go
through the full balance check regardless of whether the model belongs
to the caller's organization.
- wwwroot/v1/video/generations/index.dspy: video generation endpoint
Required params: model, llmcatelogid, prompt
Supports async task submission via existing inference infrastructure
- wwwroot/v1/image/generations/index.dspy: image generation endpoint
Required params: model, llmcatelogid, prompt
Supports both sync and async models depending on config
Both endpoints follow the same pattern as /v1/chat/completions:
1. Validate required params (model + llmcatelogid + prompt)
2. Look up llm via llm_api_map join with catalog type
3. Check customer balance
4. Route to inference (async/sync based on model config)
- Replace hardcoded colors with modern #1E293B card style
- Add 12px borderRadius to match design system
- Standardize SVG icons (36px, 1.5 stroke width)
- Fix entire_url paths to use /llmage/ module prefix
- Add page header with Title2 + description text
