AgentCreator

• New Amazon Bedrock RAG Snaps

Designer

• Concurrent pipeline viewers feature displays avatars of other users who are viewing or editing the pipeline

• Open in Monitor link and an option to download logs in the Pipeline Execution Statistics panel

Prebuilt MCP Server Tools

• Launched 545 prebuilt agentic-tool pipeline patterns that can expose 44 third-party enterprise services as MCP servers

MCP Server

• Bulk download of MCP Server logs

Monitor

• Subscription feature to extend the data available on the API Metrics page to 45 days

• The Related assets table in Asset Catalog now displays key metadata and a Last Used column for accounts and pipelines

• Revamped SnapGPT usage page featuring detailed skill-based analytics and calendar-dated trend charts

SnapGPT

• Enhanced panel in Designer with chat history and detach mode

• New feature saves AI-generated pipeline descriptions to the Info tab of a pipeline

• Added the capability to generate the agent and MCP Server pipelines using natural language prompts

The SnapLogic MCP Server

• Introduced the lease_sldb_upload tool for uploading large data files to SLDB

• Enhanced pipeline execution with custom parameters and timeouts

• Added 15 new tools for comprehensive monitoring, diagnostics, and administrative management

• Three new Amazon Bedrock Snaps that connect SnapLogic pipelines to an Amazon Bedrock Managed Knowledge Base for production RAG:

  • MKB Retrieve: Single runs a semantic search and returns matching source chunks.

  • MKB Agentic Retrieve: Performs multi-step agentic retrieval with optional answer generation and citations.

  • MKB Ingest Documents: Initiates batched ingestion (such as text, Base64 bytes, or S3 reference) into a custom data source.

• Tool pipeline timeout in Agent Snap: You can now increase the timeout to accommodate long-running tools — such as slow external APIs, large batch pipelines, and human-in-the-loop steps — to avoid a Timed out waiting for pipeline result error. Conversely, you can lower the timeout value below the default to achieve faster fail-fast behavior for tools that are stuck. The timeout applies per individual tool-pipeline execution, not cumulatively across the agent's iterations. The timeout value for existing pipelines defaults to 10 minutes.

• When you migrate an API or Proxy from Classic APIM and publish the resulting Service, the original asset becomes read-only to prevent accidental edits that would render it out of sync with the Service.

• A new subscription feature for the Monitor API Metrics page saves data for the last 45 days. Contact your CSM for more information.

• When using the Migration Tool to migrate Classic APIM assets to APIM 3.0, the premature activity log is no longer triggered before migration completes.

• Adding a new header in the Request Transformer rule where the header value evaluates to a list (for example, request.roles under API Key authentication) no longer silently ignores the header.

• In the Monitor Activity Log, activities such as creating an API/proxy, creating a service, and publishing a service/version show as generic 'Asset update' entries and do not show when filtered for APIM.

• Concurrent pipeline viewers: With this feature enabled, the page header displays avatars of other users who are viewing a pipeline and haven’t been idle for more than five minutes. This can alert users that someone else might be editing the pipeline. Contact your CSM to enable this feature. Learn more.

Open

• The Pipeline Execution Statistics panel in Designer now includes an Open in Monitor link that navigates directly to pipeline execution in Monitor. Both the pipeline name and the new Open in Monitor link are clickable hyperlinks. If the pipeline name is long, it truncates with an ellipsis, leaving the Open in Monitor link fully visible. Learn more.

  Open

  

• The Pipeline Execution Statistics panel in Designer now includes a dropdown menu for downloading standard logs or the full runtime dump without switching to Monitor.

  Open

  

• Previously, every save in the Snap configuration triggered a full validation cycle, regardless of the fields modified.

  With the September 4.45.1.0 Release, modification and save of informational fields (Label, Notes, Tags) do not trigger validation. If any configuration or functional fields are modified in the same save, full validation is done as before. This makes editing these fields faster, especially in complex pipelines where validation can introduce noticeable delays. Learn more.

The Google Sheets Snap Pack documentation has been migrated to the new documentation site and will no longer be updated in the Confluence documentation site.

• The MCP servers page (Monitor → AI Observability) now supports bulk log export. The Download logs UI button replaces one-at-a-time downloads when auditing or troubleshooting MCP traffic across many servers, pulling a whole fleet's request logs in a single action for the exact window you're investigating. To use this feature, set the time range, select servers, and click Download logs.

• A subscription feature makes data for the last 45 days available on the API Metrics page (instead of the last 72 hours). Contact your CSM for more information.

• The Asset Catalog now displays the most relevant metadata in the Related assets table when viewing pipelines that reference a specific account or pipeline. The table now features a Last Used column for both accounts and pipelines, indicating when each was last used in an execution. Learn more.

  Open

  

• Enhancements in the SnapGPT usage page:

  • Active user tiles with the exact time ranges: The KPI tiles now show Active users (Yesterday), Active users (Last 7 days), and Active users (Last 30 days), replacing the previous Daily, Weekly, and Monthly active user labels, so you can see the exact time range each value covers at a glance.

    Open

    

  • Detailed skill-level usage breakdown: The usage breakdown is now a pie chart and skill-by-skill table that shows prompt counts and percentages across all SnapGPT skills, including both the Monitor and Designer-context skills.

    Open

    

  • Calendar dates on trend charts: The Active users and Prompts charts now display actual calendar dates on the X-axis instead of relative labels (such as "7d ago"), making it easier to correlate usage spikes with specific events.

    Open

    

  • Updated UI labels: The Prompts Run tile, All actions filter, and MOST-USED action column are renamed to Prompts, All skills, and MOST-USED skill for consistency with SnapGPT terminology. Learn more.

Introduced prebuilt agentic-tool pipeline patterns that expose external enterprise services as standardized MCP servers.

• Executable from any MCP-compatible AI agent via the Pipeline Execute mechanism, these tool patterns enable the agents to perform real-time data retrieval, administrative actions, and multi-system task automation without requiring custom integration code (Zero-Code Tooling).

  • Download prebuilt tool patterns, import them into your SnapLogic environment, and expose them to AI agents as callable MCP tools.

• Over 545 prebuilt tools spanning 44 enterprise technologies are available.

• Leverage the in-house Pipeline Execute mechanism to enforce consistent authentication, error handling, and parameter passing.

• Requires an MCP Server in your environment. For setup details, refer to the MCP Server Pipeline Builder.

• The Retrieve pipeline list and Retrieve account list APIs have been enhanced to let you filter by related assets and track when each asset was last used. Pass an account_id on the pipelines endpoint to retrieve only pipelines that use a specific account, and add include_relation_details=true on either endpoint to get a last_used_at timestamp showing the most recent activity between the two assets.

• SnapCode now connects to the SnapLogic MCP server correctly on the latest Claude Code versions. A recent Claude Code update stopped forwarding plug-in credentials to the MCP connection, causing a 404 authentication error.

• SnapCode installation no longer fails on machines with a Python version newer than 3.8. If your existing install picked an incompatible interpreter, it will be corrected automatically in your next session.

• When the setup fails, SnapCode now reports the actual error. Previously, a generic "credentials missing" message appeared for unrelated errors, requesting you to re-check the configuration that was already correct.

• Generated expressions now use only functions supported by the SnapLogic expression language. Previously, constructs such as the new operator were generated but failed to compile on the platform.

• Timezone and date formatting in locally executed pipelines now match the SnapLogic Platform behavior, so local test results more accurately reflect what happens after deployment.

• On Windows, SnapCode setup now verifies environment variables in the current terminal session rather than the registry, so a correctly configured environment no longer appears broken.

• You can now generate agent and MCP Server pipelines directly from SnapGPT using natural language prompts. Learn more.

• Open

  

  The SnapGPT panel in Designer has been updated with the following new features and enhancements:

  • Chat History panel: Access all your previous conversations from the Chat History panel. Pin important chats, rename them for easy reference, remove those you no longer need, or search for them in the list, so you can pick up where you left off without re-entering a context. Learn more.

    Open

    

  • Detach mode: Float the panel anywhere on the screen and resize it to keep SnapGPT visible while working on specific areas of the canvas without the panel occupying a fixed sidebar space. Learn more.

    Open

    

  • Refreshed skills list: Skill names have been updated with short descriptions of their functionality, making it easier to identify and select the right skill for your task. Learn more.

    Open

    

• Save pipeline description to the Info tab: The describe pipeline skill now includes a Save pipeline description button that enables you to add an AI-generated description to the Notes field in the pipeline's Info tab. Learn more.

  Open

  

• Cache Service support for Ultra and ground Triggered Tasks: The Cache Service now caches two additional types of task executions. It removes a synchronous control-plane round-trip on each execution, reducing start-latency and control plane load for high-frequency runtimes or spikes in task traffic, adding resilience to sporadic control plane disruptions.
  
  

• All ServiceNow Snap Pack Snaps now process the Retry-After header value when ServiceNow returns a 429 rate-limited response.

• The documentation links in the Settings dialog for Database Snap Pack Accounts currently lead to an error page. As a workaround to view these pages, open the Snap document link and navigate to the Account page from the table of contents, the related information links for the Snap document, or the site-wide Search.

• Introduced the new lease_sldb_upload tool, enabling AI agents to upload large data files directly to SLDB for pipeline processing, without server-side size limits.

• Introduced pipeline_params and timeout parameters for the execute_pipeline tool, so AI agents can pass input values to a pipeline and set a custom wait time without any additional configuration.

• Introduced 15 new MCP Server tools that enable AI agents to go beyond running pipelines and monitor pipeline runs, diagnose failures, check environment health, inspect Snaplex versions, manage users and groups, and review scheduled tasks, without leaving the MCP session. Learn more.