Skip to main content

Documentation Index

Fetch the complete documentation index at: https://docs.elitea.ai/llms.txt

Use this file to discover all available pages before exploring further.

​
Release Overview

ELITEA 2.0.2 focuses on making day-to-day work easier to understand and easier to control. This release adds self-service publishing, clearer indexing feedback, more useful notifications, and a set of workflow improvements across model setup, pipelines, credentials, and toolkit administration.

​
Information

​
New Features

​
Self-Service Agent Publishing in Agents Studio

You can now publish agents directly to Agents Studio without depending on a separate admin handoff, which makes it faster to move production-ready agents from development into shared use. What you can do now:
  • Publish agents directly: Move approved agents into Agents Studio from the product workflow.
  • Run quality checks before publishing: Review agent readiness with built-in validation before release.
  • Manage multiple published versions: Keep more than one published version available when needed.
  • Unpublish safely: Remove an agent from new use without breaking existing conversations that already rely on it.
Benefits: Faster release cycles for agent teams, less operational overhead, and a clearer path from creation to production use. For more information, see [Agents Publishing](../how-tos/agents-pipelines/agent-publishing.mdx.

​
Indexing Completion Summary Report

Indexing now ends with a clearer completion summary so you can immediately see what was fully indexed, what was only partially processed, and what failed. What you can do now:
  • Review totals at a glance: See how many files were processed successfully, partially, or not at all.
  • Inspect file-level outcomes: Identify which specific files were skipped or failed.
  • Understand processing gaps: See the reason when content is skipped or cannot be processed.
  • Confirm indexing completion more confidently: Use the summary as a quick post-run check instead of reviewing logs manually.
Benefits: Better visibility after indexing runs, faster troubleshooting, and less ambiguity about what content is available for AI use. For more information, see Indexing Overview.

​
Personal Access Token Expiration Alerts

ELITEA now warns token owners before a personal access token expires so they have time to replace access before automations or integrations are interrupted. What you can do now:
  • Receive a 24-hour warning: Get notified before a personal access token reaches its expiration time.
  • See which token needs attention: The alert identifies the token that is about to expire.
  • Act from the notification flow: Open token management and replace the token before it stops working.
  • Avoid surprise interruptions: Catch expiring access earlier in routine work.
Benefits: Fewer unexpected authentication failures and less downtime for user-managed integrations. For more information, see Create Personal Access Token.

​
Bucket Retention Expiration Alerts

ELITEA now warns bucket owners before retention rules remove files, so they can review the bucket and keep anything important before cleanup happens. What you can do now:
  • Receive advance retention alerts: Get a warning before files are deleted by bucket retention rules.
  • See which bucket is affected: The notification points to the bucket that needs review.
  • Open the bucket directly: Navigate from the alert to the storage location and take action.
  • Prepare before cleanup runs: Move or preserve files that should not be deleted.
Benefits: Better storage governance, fewer accidental data losses, and more predictable bucket lifecycle management. For more information, see Artifacts and Notification Center.

​
Changed Features

​
Clearer Vision Capability in Model Settings

Model settings now show whether a selected model supports vision, making it easier to tell when a model can analyze image content and reducing confusion with image generation. What you can do now:
  • See vision status immediately: Check whether vision is enabled or disabled for the current model.
  • Understand what vision means: Use the inline help to confirm that vision covers reading, analyzing, and describing images.
  • Choose models more confidently: Distinguish image understanding from image generation before you run a task.
Benefits: Clearer model selection, fewer incorrect assumptions about image capabilities, and a more transparent AI configuration experience.

​
More Reliable Chat Message Metrics in Analytics

Analytics now reports chat message activity more accurately so the Overview, Agents, and Users views reflect recent conversation usage. What you can do now:
  • Track chat volume more accurately: Review updated chat-message totals in Analytics.
  • See daily activity in charts: Use the Chat Messages visualization across supported date ranges.
  • Inspect user-level usage: Confirm that user detail views reflect actual chat activity.
Benefits: More trustworthy adoption reporting and better visibility into real chat usage. For more information, see Analytics.

​
Shared Toolkit Credential Setup Is Clearer for Team Members

When a shared toolkit depends on user-specific credentials, ELITEA now gives clearer guidance about which credential ID is required and how each user should resolve missing or mismatched credentials. What you can do now:
  • See the required credential ID: Identify the exact credential each team member must create or match.
  • Follow clearer missing-credential guidance: Use structured warnings instead of ambiguous errors.
  • Jump directly into credential creation: Open a create-credential flow with the right context.
  • Avoid accidental toolkit changes: Confirm updates before altering shared toolkit credential assignments.
Benefits: Faster onboarding for shared toolkits, fewer credential mismatches, and safer collaboration in team environments. For more information, see How to Use Credentials and Creating Toolkits with User-Specific Credentials.

​
More Resilient Indexing and Clearer Error Reporting

Indexing now handles mixed-quality document sets more gracefully by continuing past unsupported files and by pointing out which file caused a failure when intervention is needed. What you can do now:
  • Keep indexing running after file-level failures: Unsupported or corrupted files no longer stop the entire job.
  • Skip problematic content automatically: Let ELITEA continue with files it can process.
  • See which file needs attention: Review clearer error messaging that identifies the file involved.
  • Finish large runs with less manual recovery: Index broad file sets without restarting because of a single bad item.
Benefits: Higher indexing success rates, faster remediation, and less time lost to avoidable reruns. For more information, see Indexing Overview and How to Index SharePoint Data.

​
Better Visibility for Cross-Page Artifact Uploads

When an uploaded file is routed to a different artifact page, ELITEA now keeps the upload status visible so you can follow progress without guessing where the file went. What you can do now:
  • Track upload progress from the current page: Stay informed even when the destination is another artifact location.
  • Understand where the file is being stored: Follow clearer cues to the correct artifact page.
  • Monitor long uploads more reliably: Keep visibility during uploads that take longer to complete.
Benefits: Less confusion during file handling and a more consistent upload experience across artifacts. For more information, see Artifacts.

​
Jira and Confluence Setup Is Simpler and Safer

Jira and Confluence configuration has been reorganized so hosting and API version choices are easier to understand, easier to reuse, and less likely to be misconfigured. What you can do now:
  • Set hosting at the credential level: Choose Auto, Cloud, or Server once in the credential instead of repeating the setting in every toolkit.
  • Use guided API version selection: Select Auto, V3, or V2 from a dropdown instead of relying on free-text input.
  • Let defaults follow your hosting setup: Use automatic version resolution based on the linked credential.
  • Work in cleaner forms: Use reorganized Jira and Confluence toolkit layouts with advanced options grouped more clearly.
Benefits: Fewer setup mistakes, more predictable behavior across shared toolkits, and a cleaner administration flow for Jira and Confluence integrations. For more information, see Jira Toolkit Guide, Confluence Toolkit Guide, and Credentials.

​
Scrollable Toolkit Chat History

Toolkit chat sessions now handle longer outputs more cleanly by keeping the chat panel scrollable instead of forcing awkward review of large result sets. What you can do now:
  • Scroll through long toolkit conversations: Review large outputs without losing your place.
  • Navigate extended responses more easily: Move through long histories with a visible scrollbar.
  • Inspect tool output with less friction: Revisit earlier responses without collapsing the workflow.
Benefits: Better readability for long toolkit runs and a more usable chat panel for troubleshooting or analysis.

​
Azure DevOps Pull Request Responses Include Descriptions

The Azure DevOps get_pull_request tool now returns pull request descriptions so you can review more of the PR context in a single response. What you can do now:
  • Read PR descriptions directly: Get the description content together with the rest of the pull request details.
  • Work with fuller review context: Use descriptions in downstream prompts and summaries.
  • Handle empty descriptions safely: Continue working even when a pull request has no description.
Benefits: Better PR context for reviews, summaries, and automation built on Azure DevOps data.

​
Notification Center Works More Like an Inbox

Notification Center now includes stronger management tools so you can search, review, and clean up notifications more efficiently. What you can do now:
  • Browse notifications in a table view: Work with a more structured list layout.
  • Search server-side: Find matching notifications faster, even in larger histories.
  • Select multiple items at once: Use checkboxes for bulk actions.
  • Manage messages in batches: Mark items as read or delete them in one step.
Benefits: Faster notification triage and a more scalable way to manage high notification volume. For more information, see Notification Center.

​
Search Suggestions Are More Relevant to the Page You Are On

Entity search now narrows tag suggestions to the current page context so you see filters that make sense for the type of item you are searching. What you can do now:
  • See agent-specific tags on agent pages: Use suggestions that match the current entity type.
  • See pipeline-specific tags on pipeline pages: Avoid unrelated filter options.
  • Find results with less cleanup: Build searches with more relevant suggestions from the start.
Benefits: Faster discovery and more precise search filtering across entity management screens. For more information, see Search Entities.

​
Pipeline Creation and Editing Are More Guided

Pipeline authoring now follows a clearer save-first workflow and adds stronger inline assistance while you configure nodes and related entities. What you can do now:
  • Create pipelines in a staged flow: Enter core metadata first, then continue configuration after the initial save.
  • Use variable suggestions in more node fields: Get autocomplete help in LLM, HITL, Printer, Code, MCP, and Toolkit inputs.
  • Refresh broken cards inline: Re-fetch toolkit, MCP, agent, or pipeline cards without reloading the page.
  • Preserve work while correcting configuration issues: Resolve warnings and stale cards without losing unsaved changes.
Benefits: Faster pipeline setup, fewer avoidable mistakes, and smoother recovery when related entities change. For more information, see Pipelines.

​
Fixed Issues

  • #4718 qTest now includes all images instead of only the last one: Multi-image test cases return descriptions for every extracted image.
  • #4717 qTest now separates text and generated image descriptions more clearly: Image descriptions are no longer appended without a usable boundary.
  • #4716 qTest now generates image descriptions even without a custom prompt: Image extraction uses the default prompt when no override is provided.
  • #4667 SharePoint OAuth authorization now works from the chat-panel Authorize button: Re-authentication in chat completes successfully instead of failing in the token-exchange step.
  • #4666 Agent-node output now saves into the configured output state variable: Child execution results are stored correctly for downstream pipeline use.
  • #4651 Agent nodes now return the child agent’s final response instead of raw history JSON: Nested agent output is displayed in a cleaner, user-facing format.
  • #4531 qTest now filters base64 images from precondition fields correctly: Embedded image data in preconditions is processed through the expected cleanup flow.
  • #4530 qTest now filters base64 images from test-step descriptions correctly: Inline image data is no longer leaked when attribute order varies inside the image tag.
  • #4509 AI Configuration now shows complete and correctly sorted model lists: Provider groups no longer stop at 10 models, and model names are ordered more consistently.
  • #4503 Agent-node runs now show nested execution chips correctly: Calls to sub-agents and sub-pipelines triggered through Agent nodes are now visible in the execution flow.
  • #4489 Nested sub-agent pipelines now run correctly in Team projects: Team-project execution no longer fails in the same pipeline-state path.
  • #4470 Analytics chat-message metrics now update correctly: Recent conversation activity is reflected more reliably across Analytics views.
  • #4430 Bucket retention updates now trigger expiration notifications correctly: Notification timing stays aligned when retention settings are changed.
  • #4388 Child pipeline calls now display their call chips in execution flow: Pipeline calls initiated from agents are represented more clearly in the UI.
  • #4380 Agent-node task fields now show variable suggestions correctly: Dropdown suggestions now appear in the Agent node task field where input mapping behavior is expected.
  • #4360 Anthropic models can now read images from artifacts correctly: Artifact-based image analysis works more reliably with supported Anthropic models.
  • #4356 Project membership conflicts now show the correct error: Adding a user who is already assigned no longer shows a false success message.
  • #4353 GitHub write actions now stay on the active branch: GitHub toolkit changes no longer default unexpectedly to a protected main branch.
  • #4352 Remote MCP toolkit creation is safer when secrets are not used: Secret values are no longer exposed in the same way during setup.
  • #4348 Later pipeline LLM nodes now still require first-use approval for sensitive tools: Earlier approvals no longer bypass authorization in subsequent pipeline stages.
  • #4346 Sensitive tools called through nested agents now require authorization: Guardrails are applied more consistently in nested execution paths.
  • #4323 SharePoint list retrieval now returns only actual lists: Folder-like results are no longer mixed into the SharePoint lists response.
  • #4192 Empty CSV files now return a consistent loader response: CSV handling behaves more predictably when there is no data to process.
  • #4140 Postman collection update tools now return the updated fields: Description, variables, and authentication changes can now be confirmed directly from the response.
  • #4131 Agent-node failures are now surfaced instead of being skipped silently: Pipeline runs show child-agent errors more clearly, which makes troubleshooting easier.
  • #3937 Cursor now handles ELITEA MCP tool responses correctly: MCP tool results no longer fail in Cursor because of the earlier response-validation problem.
  • #3871 Published agents now use published child versions at runtime: Nested published agents no longer resolve unexpectedly to a latest draft version.
  • #3804 Postman folder lookup works correctly during request creation flows: Folder-based request creation no longer fails in the same way during chained Postman actions.
  • #3761 Pipeline history no longer shows zero-second duration for non-LLM runs: Pipelines that use only toolkit, printer, MCP, or similar non-LLM nodes now show the correct execution time.

​
Known Issues

  • #4471 Nested sub-agents and sub-pipelines can show incorrect message counts: Message totals in the UI may be inaccurate even when token usage is calculated correctly.
  • #4469 In-context budget values can reset after running a pipeline from chat: Budget settings may return to an earlier state after certain pipeline execution flows.
  • #4384 Swarm Mode can keep stale context when a nested agent is attached: Some nested-agent setups may not refresh context as expected.
  • #4333 Sensitive-tool approval can still restart earlier tool execution in some agent runs: In specific multi-step flows, previous tool results may be lost after approval and the agent may repeat earlier steps.
  • #4269 Large Excel indexing can still fail during embedding requests: Very large files may still hit timeout or connection issues while embeddings are being generated.
  • #3366 SharePoint index_data fails to retrieve data from document libraries with hyphens in folder names: The SharePoint index_data tool can fail with an “incorrect folder path” error when the target document library or folder name includes hyphens.
  • #3347 Deprecated Pipeline Node Fails When Calling a Nested Pipeline: Executions can error when using the deprecated Pipeline node to invoke a nested pipeline.
  • #3341 Index tool instances can remain stuck in In progress after stopping execution: When an indexing run is stopped from chat, an agent, or a pipeline, the related Toolkit Index entry can remain stuck in In progress instead of moving to a final stopped state.
  • #3252 Input mapping fields persist when switching to tools without required/optional fields: When switching tools in MCP/Toolkit nodes, input mapping fields from the previous tool can persist and be sent unexpectedly.
  • #3151 PPT Files Fail to Read or Index from Artifacts or SharePoint: Reading or indexing PowerPoint files can raise errors instead of completing successfully or returning a clear “format not supported” message.
  • #2922 GitLab org Toolkit does not restrict access to specified repositories when only one repository is added: GitLab org toolkits may ignore a single-repository restriction and operate across all project repositories.
  • #2304 MCP Client Disconnects on macOS: MCP client disconnects from platform despite tray showing “connected” on macOS; logs show “packet queue is empty” errors and repeated disconnect/reconnect cycles.
  • #1163 MCP Only Executes Latest Version: MCP-tagged resources execute the latest version regardless of which version has the MCP tag; only the latest version is accessible via MCP.