- Fix write_router to use actual_end_user_id instead of end_user_id
- Add task status tracking via Redis in scheduler
- Expose task_id in memory write response
- Fix logging import path in scheduler
- Replace direct memory agent service calls with unified MemoryService in read endpoint
- Update query preprocessor to use new prompt format and return structured queries
- Enhance MemorySearchResult model with filtering, merging, and ID tracking capabilities
- Add intermediate outputs display for problem split, perceptual retrieval, and search results
- Fix parameter alignment and remove unused history parameter in memory agent service
The changes remove the `ConditionGroup` abstraction and flatten condition expressions directly under `ConditionBranchConfig.expressions`. This simplifies the data model and evaluation logic, eliminating redundant grouping layers while preserving all functionality. The migration logic and group-level operators are removed as they are no longer needed.
BREAKING CHANGE: `ConditionBranchConfig.expressions` now expects a flat list of `ConditionDetail` instead of `ConditionGroup`; existing configurations must be updated to use direct condition lists.
Added support for evaluating conditions on individual fields of file objects within array[file] variables. Extended variable pool to extract fields from array elements, introduced new condition models (SubVariableConditionItem, SubVariableCondition, ConditionGroup), and added ArrayFileContainsOperator to handle contains/not_contains logic with nested sub-conditions. Includes backward compatibility migration for legacy flat expressions.
- Remove redundant `response_format` injection for VOLCANO provider since it's unsupported; rely on system prompt injection instead
- Extend system prompt JSON injection logic to cover VOLCANO and tool-enabled cases universally
- Simplify model parameter construction by removing redundant `params["model_kwargs"] = model_kwargs` assignments
- Refactor `CompatibleChatOpenAI._get_request_payload` to strip `response_format` when tools are present, avoiding strict validation errors in langchain_openai
- Fix timestamp calculation order in `datetime_tool.py` to avoid integer truncation before multiplication
- Consolidate memory search services by removing separate content_search.py and perceptual_search.py
- Update model client handling in base_pipeline.py to use ModelApiKeyService for LLM client initialization
- Add new prompt files and modify existing services to support consolidated search architecture
- Refactor memory read pipeline and related services to use updated model client approach
- Use Literal['set', 'remove'] for MetadataFieldChange.action instead of str
- Simplify field_path description to reflect current schema
- Remove redundant isinstance check in extract_user_metadata_task
- Replace UserMetadata full-object overwrite with incremental MetadataFieldChange
operations (set/remove per field path)
- Convert profile.role and profile.domain from scalar strings to lists
- Remove UserMetadataBehavioralHints and knowledge_tags fields
- Update Jinja2 prompt to instruct LLM to output incremental changes
- Update extract_user_metadata_task to apply changes via deep-copy and
per-field mutation for proper SQLAlchemy change detection
- Minor lint: remove unnecessary f-string prefixes in tasks.py
- Consolidate memory search services by removing separate content_search.py and perceptual_search.py
- Update model client handling in base_pipeline.py to use ModelApiKeyService for LLM client initialization
- Add new prompt files and modify existing services to support consolidated search architecture
- Refactor memory read pipeline and related services to use updated model client approach
* release/v0.3.0: (44 commits)
Revert "fix(web): prompt editor"
fix(web): prompt editor
fix(prompt-optimizer): handle escaped quotes in JSON parsing
fix(custom-tools): remove parameter coercion in custom tool base class
fix(core): conditionally apply thinking parameters based on model support
refactor(custom-tools): coerce query and request body parameters to schema types
fix(prompt-optimizer): support list content type in prompt optimizer
refactor(memory): unify user placeholder names and harden alias sync logic
fix(rag): replace semicolon separators with newlines in Excel parser output
fix(web): Compatible with Windows whitespace
fix(memory): make PgSQL the single source of truth for user entity aliases
refactor(rag): simplify Excel parsing logic and remove redundant chunk_token_num assignment
fix(web): Hide error message when workflow node error message equals empty string
ci(wechat-notify): add Sourcery summary extraction with Qwen fallback
fix(http-request,embedding,naive): tighten form-data validation, reduce truncation length to 8000, and disable chunking for Excel
fix(web): adjust the value of End User Name
fix(http-request): support array and file variables in form-data files upload
fix(web): change http body key name
fix(web): header user name
fix(web): calculate using the filtered breadcrumbs length
...
# Conflicts:
# web/src/views/UserMemoryDetail/Neo4j.tsx
# web/src/views/UserMemoryDetail/components/EndUserProfile.tsx
# web/src/views/UserMemoryDetail/types.ts
- Add `default_free_plan.py` to define the configuration for the Community Free Plan.
- Refactor `quota_stub.py` as a unified entry point, delegating checks to `core/quota_manager`.
- Implement core logic in `quota_manager.py` to support retrieving quotas from the premium module or configuration files.
- Update `tenant_subscription_controller` to return Community Free Plan information.
- Replace hardcoded user placeholder name lists in write_tools and
user_memory_service with shared _USER_PLACEHOLDER_NAMES constant
- Filter user placeholder names during alias merging in _merge_attribute
to prevent cross-role alias contamination on non-user entities
- Use toLower() in Cypher query for case-insensitive name matching
- Change PgSQL->Neo4j alias sync condition from 'if pg_aliases' to
'if info is not None' so empty aliases correctly clear stale data
- Skip alias merging for user entities during dedup (_merge_attribute and
_merge_entities_with_aliases) to prevent dirty data from overwriting
PgSQL authoritative aliases
- Add PgSQL→Neo4j alias sync after Neo4j write in write_tools to
ensure Neo4j user entities always reflect the PgSQL source
- Remove deduped_aliases (Neo4j history) from alias sync in
extraction_orchestrator, only append newly extracted aliases to PgSQL
- Guard Neo4j MERGE cypher to preserve existing aliases for user
entities (name IN ['用户','我','User','I'])
- Fix emotion_analytics_service query to use ExtractedEntity label
and entity_type property
The form-data validation now ensures all items in the list are of type HttpFormData. Truncation length for embedding inputs is reduced from 8191 to 8000 to accommodate tokenizer differences and avoid overflow. Excel parsing now disables chunking by setting chunk_token_num to 0, aligning with intended behavior for structured file ingestion.
- Updated form-data handling to accept both single FileVariable and ArrayVariable containing FileVariable for file uploads
- Fixed HTTP client redirect handling by enabling follow_redirects=True when downloading remote files
- Adjusted config validation to correctly require list type for form-data fields instead of HttpFormData class
feat(app): support file metadata in chat messages and DSL app overwrite
- Extended chat message file objects with `name`, `size`, and `file_type` fields across app_chat_service and workflow_service
- Added ability to overwrite existing app configurations via DSL import in app_dsl_service, including type validation and config update logic for AgentConfig, MultiAgentConfig, and WorkflowConfig
- Replace storage_services/search with new read_services/memory_search structure
- Implement content_search and perceptual_search strategies
- Add query_preprocessor for search optimization
- Create memory_service as unified interface
- Update celery_app and graph_search for new architecture
- Add enums for memory operations
- Implement base_pipeline and memory_read pipeline patterns
