limit and offset are optional parameters with default values (limit=20,
offset=0). Marking them as required in the JSON Schema forces LLM clients
to always supply them, breaking calls that rely on the defaults.
Fixes#1003
- Adjusted the expanded TopK limit to a maximum of 200 candidates to enhance HNSW efficiency.
- Updated the vector retrieval query to ensure the ORDER BY expression matches the indexed expression, improving index usage.
- Implemented local settings for HNSW's ef_search and iterative_scan within a transaction to optimize candidate retrieval.
- Added fallback logic to handle unsupported GUCs gracefully, ensuring query execution continues even with older pgvector versions.
These changes improve the performance and reliability of vector retrieval in the PostgreSQL implementation.
- Updated the formatKnowledgeBaseList function to return knowledge base information as XML, improving data structure for integration.
- Modified the formatOutput function in GrepChunksTool to format search results as XML, enhancing readability and consistency.
- Adjusted the formatOutput method in KnowledgeSearchTool to output search results in XML format, providing a structured response.
- Refactored the buildOutput function in ListKnowledgeChunksTool to generate XML output, improving the clarity of chunk information.
These changes standardize output formats across multiple tools, facilitating better data handling and integration.
- Extend RetrieveEngineRegistry with byStoreID map for DB store engines
- Add StoreRegistry interface and EngineFactory function type
- Create engine factory with per-driver SDK client construction
- Load VectorStore records from DB into registry at startup
- Wire Create/Delete to dynamically register/unregister engines
- Add comprehensive unit tests for registry and service
- Updated the prompt template to clarify the task of understanding and rewriting user questions, improving the instructions for coreference resolution and ellipsis completion.
- Renamed sections for better clarity, changing "Rewriting Goals" to "Query Understanding."
- Added critical notes for intent classification regarding image and document attachments, ensuring accurate processing based on the presence of attachments.
- Improved examples to reflect the updated intent classification logic, enhancing the overall user experience in query handling.
- Added a new logger for LLM calls, enabling detailed logging of model interactions, including request and response data.
- Introduced configuration options for enabling and specifying the log directory for LLM debug logs.
- Implemented cleanup for old log files to manage disk space effectively.
- Wrapped existing chat implementations to log calls when debug logging is enabled, enhancing traceability for model interactions.
- Updated the SplitTextParentChild function to maintain a newParents slice, ensuring accurate parent indexing when splitting text into child chunks.
- This change enhances the integrity of parent-child relationships by correctly associating child chunks with their respective parents, improving overall chunk processing.
- Implemented a new method to validate the uniqueness of storage bucket names across tenants, ensuring no two tenants can use the same bucket name.
- Integrated this validation into the CreateTenant and UpdateTenant methods, enhancing data isolation and preventing potential conflicts.
- Improved error logging for better traceability when validation fails.
This update strengthens the integrity of tenant storage configurations.
- Added a check to set expandedTopK to params.TopK if it is lower, ensuring the subquery limit is not less than the final limit.
- Removed unnecessary type casting in the SQL query for embedding distance calculations, simplifying the query structure.
This update improves the accuracy of vector retrieval operations in the PostgreSQL repository.
Introduce KnowledgePostProcess task to decouple image multimodal
processing from downstream operations like summary, question
generation, and knowledge graph extraction.
This allows unifying the processing logic across documents and
images and ensures all text content, including from images,
is properly extracted before post-processing.
- Added isValidURL function to validate URLs before rendering links in the document content component, preventing invalid links from being clickable.
- Refactored timeout handling in KnowledgeBase.vue to use setTimeout instead of setInterval, improving performance and resource management.
- Ensured that timeouts are cleared appropriately to avoid memory leaks and unnecessary polling.
This update improves the robustness of URL handling and optimizes the component's performance.
- Added validation for bucket name format to ensure it meets specified criteria, returning an error message for invalid formats.
- Simplified error messages for MinIO client creation and bucket auto-creation failures, enhancing clarity for users.
- Updated success messages to reflect the correct status of bucket creation without public-read policy settings.
This update improves the robustness of the MinIO integration and enhances user feedback during storage checks.
Multiple fixes for WeKnora Lite mode (SQLite, no Redis):
1. Redis nil pointer panic on FAQ import (knowledge.go)
- Add in-memory fallback (sync.Map) for FAQ import progress tracking
- Add nil guards for running task locks and progress queries
2. seq_id not auto-incrementing in SQLite (chunk.go, tag.go)
- GORM autoIncrement on non-PK columns doesn't work in SQLite
- Add AssignChunkSeqIDs() to pre-assign seq_ids before batch insert
- Add BeforeCreate hook for KnowledgeTag
- Use Unscoped() to include soft-deleted records in MAX(seq_id) query
3. NOW() function not available in SQLite (chunk.go)
- Replace raw SQL NOW() with time.Now() or datetime('now')
- Use dialector check for raw SQL that must differ between PG and SQLite
4. FAQ KB should reject file uploads (knowledge.go)
- Add kb.Type == "faq" check in CreateKnowledgeFromFile
5. Linux/macOS build compatibility (Makefile)
- Conditionally apply macOS-only linker flag based on uname
6. Add SQLite integration tests (chunk_sqlite_test.go)
- TestCreateChunks_SQLite_SeqIDAutoAssigned
- TestCreateChunks_SQLite_SeqIDContinuesFromExisting
- TestCreateChunks_SQLite_SeqIDUniqueAcrossKBs
- TestCreateChunks_SQLite_SeqIDAfterSoftDelete
- TestKnowledgeTag_SQLite_SeqIDAutoAssigned
- TestUpdateChunk_SQLite_NoNOWError
Co-Authored-By: Claude Opus 4.6 (1M context) <noreply@anthropic.com>
- Modified the SQL clause in the KeywordsRetrieve function to utilize ParadeDB's ||| operator for improved token matching.
- Enhanced the handling of retrieval results in the fuseOrDeduplicate function to maintain original scores for keyword-only results, ensuring better performance in FAQ scenarios.
This update optimizes the keyword retrieval process and improves the accuracy of search results.
- Added WeKnoraCloudVLM struct to handle interactions with the WeKnoraCloud API, including image and text processing capabilities.
- Enhanced model configuration to include AppID and AppSecret for secure API access.
- Updated modelService to resolve WeKnoraCloud credentials and pass them to the VLM configuration.
- Implemented Predict method for sending requests to the WeKnoraCloud API and processing responses.
This update introduces a new integration with WeKnoraCloud, expanding the application's capabilities in handling multimedia content.
- Replaced references to Docreader credentials with WeKnoraCloud credentials across multiple services and handlers.
- Introduced a new CredentialsConfig structure to manage third-party provider credentials, specifically for WeKnoraCloud.
- Updated database schema to include a credentials column for storing WeKnoraCloud AppID and AppSecret.
- Enhanced methods for retrieving and utilizing WeKnoraCloud credentials, ensuring proper encryption and decryption during storage and retrieval.
This update improves the management of WeKnoraCloud credentials, streamlining the integration and enhancing security measures.
- Added tenantService to modelService for retrieving WeKnoraCloud credentials when not provided in model parameters.
- Implemented resolveWeKnoraCloudCredentials method to automatically fetch AppID and AppSecret from tenant configuration.
- Updated GetEmbeddingModel, GetEmbeddingModelForTenant, GetRerankModel, and GetChatModel methods to utilize the new credential resolution logic.
- Modified ModelParameters to encrypt and decrypt both APIKey and AppSecret for improved security during database operations.
This update streamlines credential management for WeKnoraCloud models, enhancing security and usability.
- Removed support for video file uploads across the application, including updates to the knowledge base and document processing services.
- Updated internationalization files to reflect the removal of video-related terminology, changing labels and descriptions to focus solely on audio processing.
- Adjusted file type handling to exclude video formats, ensuring users receive accurate feedback regarding supported file types.
- Enhanced error handling to provide clear messages when users attempt to upload unsupported video files.
This update streamlines the application by focusing on audio processing capabilities and improving user experience through clearer messaging.
- Changed 'Audio Processing' to 'Audio & Video' in English, Korean, Russian, and Chinese translations to reflect expanded functionality.
- Updated ASR configuration titles and descriptions to include video processing capabilities, enhancing user understanding of the feature.
- Renamed file type labels from 'Audio Files' to 'Audio & Video' across multiple languages for consistency and clarity.
- Enhanced document upload handling in the KnowledgeBase to filter unsupported file types, providing user feedback for filtered images, videos, and audio files.
This update improves the internationalization of the application and clarifies the capabilities of the ASR feature.
- Improved error handling in the attachment processing logic by providing detailed XML-like error messages for text, audio, and document processing failures.
- Updated the audio file content format to use XML tags for better structure, replacing the previous placeholder format.
This update enhances the clarity of error reporting and improves the overall structure of attachment content.
- Updated the query content to include metadata for uploaded images and attachments, improving the context provided to the assistant.
- Modified the attachment prompt format to use XML-like tags for better structure and readability, including metadata for file type and size.
- Enhanced error handling for unsupported file content extraction, providing clearer feedback to users.
This update improves the assistant's ability to process and respond to queries involving multimedia content.
- Introduced a new prompt template for document analysis, allowing the assistant to provide detailed responses based on attached documents.
- Updated the rewrite template descriptions to clarify the handling of image and document attachments, ensuring better user understanding of intent classification.
- Added a new intent type for document-only queries to improve response accuracy.
This update enhances the assistant's capabilities in processing and analyzing documents while refining the user experience in intent classification.
- Introduced a new WeKnoraCloudSettings component for managing APPID and APPSECRET.
- Added functionality to save WeKnoraCloud credentials without automatically creating models.
- Implemented credential status checks across various components to provide user feedback on configuration status.
- Updated API endpoints to handle credential saving and status retrieval.
- Enhanced UI elements to guide users in configuring WeKnoraCloud settings.
This update improves the integration of WeKnoraCloud by allowing users to manage their credentials more effectively and receive real-time feedback on their configuration status.
Wire VectorStoreService to HTTP with 8 endpoints: types metadata, CRUD
(create/list/get/update/delete), and connection testing (raw + by ID).
Register routes, DI container bindings, and add API documentation.
Remove the hardcoded TruncatePromptTokens: 512 limit from WeKnoraCloud
BatchEmbed to allow the API to use its default truncation behavior,
avoiding unexpected truncation of longer input texts.
Co-Authored-By: Claude Opus 4.6 (1M context) <noreply@anthropic.com>
Add VectorStoreService with CRUD validation, duplicate checking (DB + env
stores), and engine-specific health checks for 6 vector database types.
Include VectorStoreResponse DTO, env store builder, engine type metadata,
and comprehensive unit tests.
Allow configuring custom WebSocket and API base URLs for WeCom
via optional credentials fields (ws_endpoint, api_base_url), enabling
integration with privately deployed WeCom instances.
Changes:
- LongConnClient and WebhookAdapter accept custom endpoint URLs
- Endpoint URLs validated at construction (scheme + SSRF check)
- Private deployment hostnames added to file download SSRF allowlist
- Frontend forms added for both WebSocket and Webhook modes
- i18n support for zh-CN, en-US, ru-RU, ko-KR
Uses default public cloud endpoints when not specified.
- Remove infrastructure/crypto package (CryptoService, config, PBKDF2-based encryption)
- Replace with lightweight utils.GetAESKey/EncryptAESGCM/DecryptAESGCM helpers
- Remove cryptoSvc dependency from modelService, tenantService, weKnoraCloudService
- Remove crypto state persistence logic from container initialization
- Add TruncatePromptTokens field to WeKnoraCloud embed request
- Update frontend hint text for WeKnoraCloud credential errors
Co-Authored-By: Claude Opus 4.6 (1M context) <noreply@anthropic.com>
draix