> For the complete documentation index, see [llms.txt](https://docs.ovaledge.com/llms.txt). Markdown versions of documentation pages are available by appending `.md` to page URLs; this page is available as [Markdown](https://docs.ovaledge.com/release8.2/service-releases/release8.1.x/release8.1.1.md). # Release8.1.1 OvalEdge **Release8.1.1** is a **service release,** introduces significant enhancements to askEdgi, focusing on governed AI-driven analytics, reusable recipe-based workflows, workspace lifecycle management, and data quality remediation. This release adds parameterized recipes, a new recipe execution framework, natural language code editing, AI data-sharing controls, expanded RAG coverage, and legacy data cleanup capabilities. The release also improves workspace governance, recipe management, and metadata-driven analysis. **Key Highlights** * **askEdgi** * Introduced Recipe as a Product (RaaP) with **parameterized recipes**, enabling reusable and governed analytical workflows across multiple business scenarios. * Enhanced workspace data management with automatic **dataset reload**, improved retention controls, and protection for workspace-generated assets. * Added **@Mention-based contextual references** for datasets, glossary terms, data products, files, live tables, and other governed assets. * Added Thumbs Up / Thumbs Down **feedback** capabilities to capture user feedback and support metadata improvement initiatives. * Introduced **AI Data Sharing** Controls to govern the use of physical data in external AI-powered analysis and enrichment workflows. * Implemented a system-wide workspace **purge** policy to improve storage management while preserving metadata, conversations, and object references. * Added **automated AI generation** of Recipe names, descriptions, and step descriptions to simplify recipe creation. * Enabled automatic persistence of recipe-generated tables with dedicated **workspace folders** and improved output traceability. * Introduced **Natural Language Code Editor**, allowing users to modify analytical logic using plain language instead of editing SQL or Python code directly. * Added **Data Cleanup** Execution capabilities to resolve legacy data quality issues directly from askEdgi workspaces. * Added a new **Recipe Execution Framework** with inline execution, execution history, dependency validation, and output review within askEdgi and Studio. * Expanded RAG coverage across additional OvalEdge assets, including Data Products, Reports, APIs, Projects, Columns, Data Stories, and governance objects. * Improved asset discovery accuracy using column-level Business Glossary associations and glossary-driven query generation. * Strengthened the agent framework architecture with specialized planning, discovery, and coding agents for improved analysis accuracy and reliability. * Revamped the Recipes experience and My Recipes pages for Enterprise and Marketplace users with simplified navigation, management, and publishing workflows. * Improved workspace object status validation to ensure assets are marked as added only after successful processing and workspace synchronization. * **Data Quality** * **Data Quality Dashboard** – All Data Quality Service Requests associated with an object, including manually reported issues, are now displayed in the Object Dashboard, providing a consolidated view of data quality remediation activities.. ## Release Details
Release TypeRelease VersionBuild <Release. Build Number. Release Stamp>Build Date
Service ReleaseRelease8.1.1Release8.1.1.8110.e698b9825 June, 2026
## askEdgi ### New & Improved ### Parameterized Recipe Creation askEdgi now supports Parameterized Recipes, enabling users to create reusable workflow templates that can be executed with different inputs at runtime. Previously, Recipes were created as static workflows with hardcoded values, requiring users to recreate the same Recipe for different datasets, filters, or business scenarios. The following enhancements have been implemented: * Users can create reusable Recipes with configurable runtime parameters. * Supported parameter types include Single Select, Multi Select, Text, Number, and Date. * Parameters can be configured with display labels, default values, required settings, and dataset-based value mappings. * Dropdown parameter values can be dynamically populated from Recipe ingredients and dataset columns. * A new AI-powered Detect & Bind Parameters capability automatically identifies hardcoded values in SQL and Python code and converts them into reusable parameters. * Recipes support multi-step workflows using SQL, Python, and AI Analysis steps. * Parameter references can be injected into execution logic to dynamically customize Recipe behavior at runtime. * Validation checks ensure parameter uniqueness, data type compatibility, ingredient availability, and valid parameter references. * Dependency management prevents deletion of ingredients, parameters, or steps that are actively used within the Recipe workflow. * Execution forms are automatically generated based on configured parameters, allowing users to provide inputs without modifying code. * Governance-approved workflows can now be reused across multiple datasets, departments, business domains, and time periods. ### Natural Language Code Editor In askEdgi, users can generate analyses and review the underlying SQL or Python code created from their prompts. Previously, modifying generated code required direct code editing, which could be challenging for business users and non-technical users. In addition, when code updates failed, error messages were not always clearly displayed in the user interface. To simplify this experience, askEdgi now introduces a Natural Language Code Editor that lets users modify analysis logic in plain language rather than editing code directly. The following enhancements have been implemented: * The existing Code Explanation view has been replaced with a Data Flow Logic view. * Data Flow Logic presents the generated SQL or Python logic as clear, step-by-step business instructions. * Users can modify the analytical logic using natural language without writing or understanding code. * askEdgi automatically regenerates the underlying SQL or Python code based on the updated logic. * Users can review the regenerated logic before execution. * Validation and execution errors are now displayed directly in the user interface with accurate and actionable messages. * When the updated analysis is executed, the previous output is replaced with the newly generated results. * The feature enables iterative refinement of analyses through conversational interactions rather than manual code editing. * Business users can adjust filters, calculations, transformations, and analytical steps without leaving askEdgi or relying on external coding assistance. * The workflow improves accessibility, reduces dependency on technical skills, and accelerates analysis refinement within the same conversation. ### Context References Using @Mentions In askEdgi, users can now explicitly reference contextual assets using @mentions to improve analysis and execution accuracy. Previously, the analysis relied primarily on automatic RAG-based context discovery, which could lead to incorrect dataset selection or misinterpretation of business logic. The following enhancements have been implemented: * Users can reference Workspace Objects, Data Catalog Tables, Files, Business Glossary Terms, Tags, Live Tables, and Data Products using @mentions. * Workspace objects are reused directly without requiring re-import. * Catalog datasets and Data Products require confirmation before being added to the workspace. * Business Glossary Terms are treated as business logic definitions during execution. * Tags function as dataset selectors and cannot be executed directly as datasets. * Live Tables execute directly against source systems without importing data into the workspace. * To maintain execution consistency, askEdgi does not support analysis across Live Tables and imported workspace datasets in the same request. Live Tables query data directly from the source system, while imported datasets use data stored in the workspace. * To analyze data from both sources, first import the required Live Table into the workspace or use datasets already available in the same environment. * The @mention search experience now includes grouped results, filter chips, object-type indicators, and access-controlled visibility. ### Thumbs Up and Thumbs Down Feedback for RAG Responses In the askEdgi module, a feedback mechanism has been introduced to capture user sentiment on generated responses and improve metadata curation workflows. The enhancement helps identify gaps in metadata quality, glossary definitions, semantic mappings, and contextual relevance that affect RAG-based analysis outcomes. The following capabilities have been introduced: * Users can now provide Thumbs Up or Thumbs Down feedback for askEdgi-generated responses. * A mandatory feedback pop-up appears when users select Thumbs Down. * Users can specify whether the issue is related to incorrect tables, incorrect interpretation, or missing context. * Additional comments can be provided to describe missing or inaccurate information. * Captured feedback is stored along with the original prompt, suggested assets, result metadata, and user comments. ### Auto Persist Recipe Generated Tables into Recipe Output Folder In the askEdgi workspace, recipe-generated tables are now automatically organized and stored for easier reuse and visibility. Previously, temporary and intermediate tables created during recipe execution were not systematically retained in the workspace. The following enhancements have been implemented: * Dedicated workspace folders named Recipe Ingredients and Recipe Output are automatically created during recipe execution. * Intermediate and final output tables are automatically stored in the corresponding folders. * Generated tables are tagged with recipe metadata for traceability. * Users can open, query, and reuse generated tables directly from the workspace. * Storage operations execute without interrupting recipe execution workflows. * Existing workspace access controls and storage policies continue to apply. ### System-Wide Purge Policy for Workspace Data and Results askEdgi now supports an automated purge policy and reload framework to improve storage management and workspace lifecycle governance. The following enhancements have been implemented: * New system settings are available to configure workspace data retention and temporary results retention periods: * askedgi.workspace.data.retention\_days – Controls how long workspace datasets are retained before being purged. The default value is set to 30 days. * askedgi.workspace.results.retention\_days – Controls how long temporary analysis results and intermediate files are retained. The default value is set to 5 days. * Imported workspace datasets are automatically purged after the configured retention period to reduce storage usage. * Cataloged Dataset metadata remains available in the ovaledge even after the underlying data in workspace is purged. * Purged datasets are marked with status indicators and can be reloaded directly from their source when supported. * A new Reload action allows users to restore purged datasets directly from the Data Catalog without re-importing the data manually. * User permissions are validated before reload operations are executed. * Reload actions now display processing indicators, prevent duplicate requests, and automatically update the workspace status when the reload completes. * Chat history and conversations are retained independently and are not affected by workspace data purging. * Clear messages are displayed when users attempt to access or perform actions on purged datasets. * To prevent accidental data loss, the automated purge policy excludes: * Local file uploads (CSV, XLSX, JSON, and similar files) * AI-generated analysis outputs * Recipe-generated outputs * Live connection datasets * These objects remain available in the workspace and do not display purge or reload options because they cannot be restored from a catalog source. * In addition, temporary analysis results, intermediate processing files, and chat-generated files are automatically removed after the configured retention period to optimize storage and maintain workspace performance. * Storage cleanup is performed automatically without affecting saved workspace metadata or conversation history. * Retention management helps reduce storage consumption while maintaining a consistent user experience. ### Data Cleanup Execution for Legacy Data Quality Issues In the askEdgi workspace, users can now clean datasets affected by legacy data quality issues directly from the workspace. When an object associated with legacy data quality issues is added to the workspace, a Data Quality Debt icon appears next to the object, allowing users to view available cleanup contexts derived from related service requests. The following enhancements have been implemented: * Users can initiate data cleanup directly using the Clean Data option. * Data cleanup execution starts in a new thread and updates the workspace with cleaned data after successful completion. * Source system data remains unchanged during cleanup execution. * Generated cleanup execution code is available for review in the created thread. ### Data Reload Enhancement for Workspace In the askEdgi workspace, data retention and reload management have been improved to prevent accidental loss of workspace-created data and to simplify the recovery of purged datasets. The following enhancements have been implemented: * Datasets purged after the configured retention period can now be reloaded directly using saved catalog references. * A new Reload option is available directly in the workspace. * Local file uploads, AI-generated outputs, Recipe-generated outputs, and Live connection datasets are retained and excluded from automated purge policies because they lack a catalog source for reload. * Uploaded files continue to require manual re-upload if intentionally removed. * Additional workspace details, including Last Imported timestamp, source connection, and purge status, are now displayed. * Missing datasets are automatically reloaded during Recipe execution before SQL or Python steps execute. * User access permissions are validated before reload operations begin. * Recipe save operations now support validation-based saving without triggering additional AI generation. * Existing content is preserved if AI generation fails during Recipe save operations. ### AI Data Sharing Control askEdgi now includes a global setting named share.data.ai to control whether physical data can be shared with external AI services. The setting is enabled (TRUE) by default after the upgrade. The following enhancements have been implemented: * AI data sharing can be enabled or disabled at the system level. * When enabled, askEdgi supports AI Enrichment, Data Quality Cleanup, logical verification, and sample data-based analysis. * When disabled, askEdgi operates only with metadata and prevents sharing of physical row data, sample values, and profiling statistics. * AI-assisted features dependent on physical data are automatically disabled when sharing is restricted. * RAG context generation, conversational memory, analysis-agent requests, and suggestion workflows respect the configured sharing policy. * Recipe execution pauses automatically for Recipes containing AI Enrichment steps when data sharing is disabled. * Restricted AI features display administrator-controlled lock messages. ### Auto AI Generation for Recipe and Step Descriptions In askEdgi Recipes, Recipe Descriptions and Step Descriptions are now generated automatically during Recipe creation. Previously, users had to manually provide these details while saving chat outputs as Recipes. The following enhancements have been implemented: * Recipe Descriptions are automatically generated when the Save as Recipe window is opened. * Step Descriptions are automatically generated for each recipe step based on the underlying analytical logic. * Users can regenerate Recipe Descriptions and Step Descriptions directly from the Recipe creation interface. * Description fields remain fully editable, allowing users to refine or customize the generated content as needed. * Regenerate actions are temporarily disabled while AI generation is in progress to prevent duplicate requests. * Recipe save operations use the content currently available, whether AI-generated, user-edited, or manually entered, without triggering additional AI generation. ### New Recipe Execution Workflow askEdgi now includes a redesigned Recipe Execution Framework that supports direct Recipe execution from the Recipe module and askEdgi Studio without leaving the current workflow. The following enhancements have been implemented: * Recipes can be executed through a right-side shutter panel within the existing workflow. * Users can review Recipe details, ingredients, execution steps, and execution history in a single interface. * Before execution, askEdgi verifies that users have access to all required Recipe ingredients (datasets and objects) and prompts them to resolve any missing access or selection issues before execution can proceed. * Previous executions can be re-run. * Outputs for visible and hidden steps can be reviewed directly within the execution flow. * Recipe execution is integrated into askEdgi Studio to support inline execution and continued analysis within the same chat experience. ### Workspace Object Addition Status Validation In the askEdgi RAG workflow, the object addition process has been improved to ensure accurate updates to workspace status during asset import and synchronization. Previously, the Added status appeared immediately after users selected the Add option, even when the object had not been fully added to the workspace. The following improvements have been implemented: * The Added status is displayed only after the object is successfully added to the workspace and all validations are completed. * Objects that fail processing or are only partially added are no longer marked as successfully added. * Status updates now accurately reflect the actual state of the object addition process. ### Business Glossary Formula Resolution During Query Generation In the askEdgi RAG workflow, an issue where Business Glossary definitions and formulas were not applied during query resolution has been resolved. Previously, when users searched with glossary-defined business terms, askEdgi generated generic SQL queries based on available columns rather than using the configured glossary business logic. The following improvements have been implemented: * Business Glossary definitions and formulas are now prioritized during query resolution when matching glossary terms are identified. * Glossary business logic is now used to generate SQL queries rather than relying solely on inferred column calculations. * Glossary terms are mapped to relevant columns and associated datasets during execution planning. * Query generation now respects glossary-defined semantic relationships for business calculations. * For glossary-defined metrics such as Total Revenue, askEdgi now derives calculations using the configured business formula instead of generic aggregate values. * The enhancement improves consistency between business terminology and generated analytical logic. * RAG-driven analysis workflows now produce more accurate and governance-aligned query outputs. * Existing Business Glossary authoring and metadata management workflows remain unchanged. ### Expanded RAG Coverage Across OvalEdge Objects In the askEdgi RAG framework, support has been expanded to include additional OvalEdge object types during discovery, context generation, and analysis planning. This enhancement improves knowledge coverage and enables askEdgi to retrieve more relevant business and technical context across the platform. **Newly Supported Object Types**
Object TypeDescription
Data Products Discover and analyze governed data products.
Data Stories Include business narratives and insights in context generation.
SchemasUtilize schema-level metadata during discovery and analysis.
ReportsRetrieve report metadata for improved business context.
APIsInclude API metadata and related information in RAG responses.
CodesLeverage code object metadata during context generation.
Projects Incorporate project-level information into discovery workflows.
Project TasksUse task-related metadata for contextual analysis.
ColumnsInclude column-level metadata to improve asset identification and relevance.
Data Quality*Utilize data quality information where applicable.
Service Desk*Include service requests and related governance context where applicable.
\* Available for contextual discovery and analysis scenarios where relevant. The following enhancements have been implemented: * Additional governance and operational objects, including Data Quality and Service Desk entities, are now available for contextual discovery where applicable. * askEdgi can now retrieve broader metadata context across supported object types to improve asset discovery and analysis accuracy. * Object relationships and metadata references are now included during context generation to improve semantic understanding. * Expanded object coverage improves the relevance of responses for business, governance, stewardship, and analytical use cases. ### Enhanced Agent Framework for Improved Analysis Accuracy The askEdgi agent framework has been enhanced to improve how analysis requests are planned, discovered, and executed. Previously, multiple tasks were processed within a shared context, which could lead to context overload and less reliable results. With this enhancement, Planning, Discovery, and Coding agents now operate with clearer responsibilities and improved context management. The following enhancements have been implemented: * Planning, Discovery, and Coding agents now process requests independently with dedicated responsibilities. * Complex requests are automatically broken into smaller tasks to improve execution accuracy. * Enhanced context management reduces information overload and improves response quality. * Specialized sub-agents can be used for activities such as asset discovery, query planning, and code generation. * Long-running and parallel tasks can be processed more efficiently through isolated agent workflows. * Reusable skills, domain knowledge, and memory are loaded only when required to improve relevance and reduce prompt complexity. * Built-in validation and verification workflows help reduce hallucinations and improve output reliability. * Context summarization and intelligent memory management improve performance during extended analysis sessions. * Human approval checkpoints can be incorporated into critical decision-making workflows where required. ### Recipes and Recipe Creation Experience Revamp The askEdgi Recipes module now provides a simplified and more intuitive experience for creating, editing, managing, and publishing recipes. The redesign improves navigation, reduces complexity, and makes recipe workflows easier for both business and technical users. The following enhancements have been implemented: Simplified recipe creation and editing workflows. * Improved navigation across recipe management pages. * Redesigned the My Recipes page with a cleaner and more organized layout. * Easier access to recipe actions and configurations. * Improved recipe discovery and management experience. * Enhanced step management and recipe maintenance capabilities. * Consistent user experience across recipe creation, editing, and publication workflows. * Reduced technical complexity to improve usability for business users. ### My Recipes Revamp for Enterprise Instance The My Recipes page in Enterprise instances has been redesigned to simplify recipe management and organizational sharing. The following enhancements have been implemented: * Redesigned the My Recipes page with an improved user interface. * Simplified recipe discovery and management workflows. * Improved process for publishing recipes to the organization library. * Easier access to organizational sharing actions. * Consistent user experience across recipe management workflows. * Enhanced usability for managing and sharing enterprise recipes. ### My Recipes Revamp for Public Instance The My Recipes page in the Recipe Marketplace has been redesigned to improve recipe management and Marketplace publishing. The following enhancements have been implemented: * Redesigned the My Recipes page with an improved user interface. * Simplified recipe discovery and management workflows. * Improved process for publishing recipes to the Marketplace. * Streamlined publication workflow for Marketplace submissions. * Consistent design across recipe-related pages. * Enhanced usability for managing and publishing recipes. These enhancements provide a more streamlined experience for creating, managing, sharing, and publishing recipes across Enterprise and Public instances. ### Fixed ### Missing Navigation Link for Created Service Requests In askEdgi, an issue where Service Request ticket IDs appeared as plain text in the Action Results table has been resolved. The following issue has been resolved: * Created Service Request ticket IDs now appear as clickable navigation links. * Users can directly open created service requests from the Action Results section after execution. ## Data Quality ### Improved ### Display All Service Requests in the Object Dashboard The Data Quality Dashboard has been enhanced to provide complete visibility into service requests associated with an object. Previously, only automatically generated Data Quality Rule service requests were displayed from the Object Summary page. With this improvement, the dashboard now displays all Data Quality Service Requests associated with the selected object, including manually reported issues. This provides a consolidated view of data quality issues and remediation activities directly from the object dashboard. ## Change Management Change Management helps stakeholders understand what has changed from the previous version to the latest version, who is impacted, and how processes or permissions are affected. This ensures smooth adoption and reduces operational risk. The impacted modules include askEdgi and Data Quality. ### Data Quality Dashboard * The Data Quality Dashboard now displays all service requests associated with an object, providing a complete view of data quality issues and remediation activities.

What Changed

In the Data Quality Dashboard, previously, only service requests generated from Data Quality Rules were displayed. Manually created service requests were not included. Now, all service requests associated with an object, including manually created requests, are displayed in the dashboard.
Affected Users: All Users

👉 For more details, see Display All Service Requests in the Object Dashboard.

### askEdgi * Administrators can now control whether workspace data can be shared with AI services using the share.data.ai setting.

What Changed

In askEdgi, previously, AI-powered features that required workspace data processing were available without centralized control over data sharing. Now, administrators can use the share.data.ai setting to control AI-powered capabilities across askEdgi.
Affected Users: All Users

👉 For more details, AI Data Sharing Control.

* Recipe-generated tables are now automatically saved in the Recipe Output folder for easier access and reuse.

What Changed

In askEdgi, previously, recipe-generated outputs were available only during recipe execution and were not systematically stored in the workspace. Now, all intermediate and final output tables generated during recipe execution are automatically saved in the Recipe Output folder.
Affected Users: All Users

👉 For more details, Auto Persist Recipe Generated Tables into Recipe Output Folder.

* Workspace data is now managed using configurable retention policies with support for on-demand data reload.

What Changed

In askEdgi, previously, workspace data remained available until it was manually removed. Now, datasets are automatically purged after the configured retention period. Metadata, catalog references, and chat history remain available after the data is removed. When a dataset is purged, users can use the Reload option to restore the dataset from its source and continue their analysis.
Affected Users: All Users

👉 For more details, Data Reload Enhancement for Workspace.

* Users can now reference governed assets directly in prompts using @mentions.

What Changed

In askEdgi, previously, the system relied on prompt interpretation and retrieval processes to identify relevant datasets, glossary terms, and other assets. Now, users can type @ and directly select supported objects such as catalog assets, business glossary terms, data products, tags, live sources, and workspace assets. askEdgi uses the selected objects as the analysis context, improving accuracy and reducing execution time.

Affected Users: All Users

👉 For more details, Context References Using @Mentions.

* Recipe execution now provides a cleaner and more organized user experience.

What Changed

In askEdgi, previously, recipe execution displayed all execution steps and intermediate outputs directly in the chat, making it difficult to focus on the final result. Now, recipe execution displays only the final output by default while showing the progress of each step during execution. Users can expand execution details to review intermediate steps and outputs when needed. Recipe-generated outputs are also automatically saved in the workspace for future use.

Affected Users: All Users

👉 For more details, New Recipe Execution Workflow.

### **Advanced Jobs** {% hint style="info" %} There are no new Advanced Jobs or updates to existing Advanced Jobs in this release. {% endhint %} ## System Settings This release introduces new system settings that enhance user control over the application's behavior.
KeyDescriptionImpacted Modules
embedding.enable

Controls whether document embeddings are generated and populated for catalog search. It determines whether semantic search capabilities are available, enabling users to discover relevant catalog assets based on contextual meaning rather than exact keyword matches.

Parameters:

  • When enabled, document embeddings are generated and populated for catalog search, allowing semantic search functionality.
  • When disabled, document embeddings are not generated, and catalog search relies on standard search mechanisms.
  • Default: False
AI
embedding.config

Defines the configuration parameters used for catalog embedding generation. It controls token limits, chunking behavior, overlap settings, processing parallelism, and batch sizes, helping to optimize embedding creation for semantic search and other AI-driven catalog capabilities.

Parameters:

  • EMBEDDING_MODEL_MAX_CONTEXT_TOKENS: Specifies the maximum number of tokens supported by the embedding model for a single request.
  • EMBEDDING_INPUT_TOKEN_HEADROOM: Reserves a portion of the model's token limit to prevent requests from exceeding the maximum context size.
  • EMBEDDING_API_CHARS_PER_TOKEN: Defines the character-to-token conversion ratio used when estimating token counts for content truncation.
  • MAX_TOKENS_PER_CHUNK: Specifies the maximum number of tokens allowed in each content chunk before embedding generation.
  • CHUNK_OVERLAP_TOKENS: Defines the number of tokens shared between adjacent chunks to preserve context across chunk boundaries.
  • BULK_EMBEDDING_PARALLELISM: Specifies the number of embedding generation processes that can run concurrently during bulk operations.
  • EMBEDDING_BATCH_SIZE: Defines the number of content chunks processed in a single embedding generation batch.
AI
### Existing System Settings - Updated This release includes updates to existing system settings, enhancing functionality to improve data-processing accuracy and operational efficiency across supported modules.
NameUpdated FieldUpdated Field Value
ai.configValue

JSON_SET(

CAST(jobparmvalue AS JSON),

'$.providers.anthropic',

JSON_OBJECT(

'classification_model', 'claude-sonnet-4-20250514',

'summarization_model', 'claude-sonnet-4-20250514',

'reasoning_model', 'claude-3-opus-20240229',

'large_context_model', 'claude-3-opus-20240229',

'api_key', '',

'advanced_settings', JSON_OBJECT(

'base_url', 'https://api.anthropic.com',

'timeout', '60',

'temperature', '0.7',

'top_p', '0.8',

'top_k', '40',

'max_tokens', '4096',

'retry_count', '3',

'stop_sequences', ''

)

),

'$.providers.bedrock',

JSON_OBJECT(

'classification_model', 'us.anthropic.claude-3-5-haiku-20241022-v1:0',

'summarization_model', 'us.anthropic.claude-sonnet-4-5-20250929-v1:0',

'reasoning_model', 'us.anthropic.claude-sonnet-4-5-20250929-v1:0',

'large_context_model', 'us.anthropic.claude-sonnet-4-5-20250929-v1:0',

'advanced_settings', JSON_OBJECT(

'region', 'us-east-1',

'access_key_id', '',

'secret_access_key', '',

'session_token', '',

'profile', '',

'role_arn', '',

'role_session_name', 'ovaledge-bedrock',

'timeout', '120',

'temperature', '0.7',

'top_p', '',

'max_tokens', '4096',

'retry_count', '3',

'embedding_model', 'amazon.titan-embed-text-v2:0_1024'

)

)

)

*** Copyright © 2026, OvalEdge LLC, Peachtree Corners, GA USA --- # Agent Instructions This documentation is published with GitBook. GitBook is the documentation platform designed so that both humans and AI agents can read, navigate, and reason over technical content effectively. Learn more at gitbook.com. ## Querying This Documentation If you need additional information that is not directly available in this page, you can query the documentation dynamically by asking a question. Perform an HTTP GET request on the current page URL with the `ask` query parameter, and the optional `goal` query parameter: ``` GET https://docs.ovaledge.com/release8.2/service-releases/release8.1.x/release8.1.1.md?ask=&goal= ``` `ask` is the immediate question: it should be specific, self-contained, and written in natural language. `goal` is optional and describes the broader end goal you are ultimately trying to accomplish on behalf of the user. GitBook uses it to tailor the answer towards what is most useful for that goal. The response will contain a direct answer to the question and relevant excerpts and sources from the documentation. Use this mechanism when the answer is not explicitly present in the current page, you need clarification or additional context, or you want to retrieve related documentation sections.