feat(conversation): hard-cut text2sql v2 workflow spine and RAG runtime gates#23
Merged
Merged
Conversation
- Added `AGENT_RAG_RETRIEVAL_ENABLED` to the environment configuration, allowing for toggling of RAG retrieval functionality. - Updated `AppConfigService` to include a method for retrieving the new RAG retrieval setting. - Enhanced `RetrieveKnowledgeNode` to respect the RAG retrieval configuration, providing a fallback mechanism when retrieval is disabled. - Removed unused `GraphService` and related components to streamline the codebase and improve maintainability.
- Updated the `generatedAt` timestamps in gate summary JSON files for relationship platform, semantic spine, and modeling parity shadows to reflect the latest generation time. - Changed `strictMode` from `true` to `false` in the modeling parity shadow gate summary to allow for more flexible processing. - Enhanced the semantic plan service to handle degraded contexts without grounding evidence, ensuring the SQL path is maintained in such scenarios. - Introduced a tolerance for gate comparison in the text2sql evaluation service to prevent false negatives due to rounding errors.
- Added new endpoints for managing RAG task configurations, including listing, upserting, and health checking. - Implemented a new `RagTaskConfig` model and service to handle RAG task settings. - Enhanced the `SettingsService` to integrate RAG task configuration management. - Updated the `EmbeddingRouterService` and related components to utilize RAG task configurations for embedding runtime resolution. - Introduced coverage gap handling in the semantic plan service to improve error reporting and decision-making in SQL generation.
- Introduced new evaluation gates for Text2SQL v2, including strict modes for evaluation and focused coverage. - Updated CI references in documentation to reflect new gate commands and their expected outputs. - Enhanced the text2sql evaluation service to include detailed traceability metrics, ensuring comprehensive coverage reporting. - Added new traceability fields in evaluation cases to track fixture families, behavior tests, and flow nodes. - Improved error handling in SQL generation to account for metadata and general queries, preventing unnecessary SQL execution. - Updated test fixtures to include traceability information for better integration testing.
- Introduced a new method `runOperationalFailure` in `FormatAnswerNode` to handle operational failures with user guidance. - Updated `RetrieveKnowledgeNode` to support datasource and allowed tables in its input, improving context handling. - Enhanced `SqlGenerationService` to utilize semantic shortcuts for generating SQL queries, improving efficiency and error recovery. - Implemented checks in `SqlReadonlyTool` to prevent introspection of schema system tables, ensuring security compliance. - Added new tests for RAG provider model previews and SQL generation shortcuts, ensuring robust functionality and error handling.
Introduce trace.v2 runtimePlan, artifactRefs, and Smart Defaults evidence across the shared contract, LangGraph runtime, delivery projection, and stream state summaries.
Extend focused and eval gates so runtime intelligence requires producer-backed artifact refs, stream lifecycle coverage, Smart Defaults evidence, and no-legacy owner scans.
Define additive plan ledger obligations, generation claims, validation fulfillment, and delivery summary fields so the v2 trace can carry SQL intent contracts without changing runtime plan semantics.
Build semantic plan obligations, block failed hard blockers before SQL generation, record fulfillment claims on generated SQL, and validate SQL against the ledger before execution or correction.
Add ledger fixture families, rollout metrics, and unit coverage for obligation creation, generation claims, validation fulfillment, correction grounding, and safe delivery surfaces.
This file contains hidden or bidirectional Unicode text that may be interpreted or compiled differently than what appears below. To review, open the file in an editor that reveals hidden Unicode characters.
Learn more about bidirectional Unicode characters
Sign up for free
to join this conversation on GitHub.
Already have an account?
Sign in to comment
1 participant
Add this suggestion to a batch that can be applied as a single commit.This suggestion is invalid because no changes were made to the code.Suggestions cannot be applied while the pull request is closed.Suggestions cannot be applied while viewing a subset of changes.Only one suggestion per line can be applied in a batch.Add this suggestion to a batch that can be applied as a single commit.Applying suggestions on deleted lines is not supported.You must change the existing code in this line in order to create a valid suggestion.Outdated suggestions cannot be applied.This suggestion has been applied or marked resolved.Suggestions cannot be applied from pending reviews.Suggestions cannot be applied on multi-line comments.Suggestions cannot be applied while the pull request is queued to merge.Suggestion cannot be applied right now. Please check back later.
Summary
Testing