🌷 Late Spring 2026 Platform Update: Nexadata MCP Server, Runtime Variables & Enhanced Excel Support

Hi there,

This update introduces the Nexadata MCP Server, a major new surface that connects the Nexadata platform to the broader ecosystem of agentic AI clients through the Model Context Protocol. Alongside MCP, this release delivers Runtime Variables for parameter-driven workflows, Execution Statistics across both the platform UI and the MCP Server, and Enhanced Excel Support for semi-structured data sources.

This is a foundational release for agentic data integration. The MCP Server, combined with runtime variables and execution telemetry, makes Nexadata a first-class participant in multi-agent automation workflows.

🔗 Nexadata MCP Server

The Nexadata MCP Server is now available, exposing Nexadata workflows, pipelines, and execution context to any MCP-compatible client. The server implements the open Model Context Protocol, the emerging standard for connecting LLMs and AI agents to real-world systems.

This means an MCP-compatible client can connect directly to a Nexadata environment, reason against live platform state, and take action on behalf of the user without leaving its native interface.

What the MCP Server Exposes

The MCP Server provides tools for:

• Workflow and pipeline discovery: List, inspect, and search across workflows and supporting or output pipelines in a workspace
• Run state and history: Retrieve the status of recent runs, including pass/fail outcomes and timing information
• Edit attribution and audit history: Surface "who made the last change and when" for pipelines and mapping groups, giving the LLM context on recent platform activity
• Execution telemetry: Provide real-time run statistics for diagnostic conversations (covered in detail in the Execution Statistics section below)

Reasoning Context for Warnings and Errors

A core design goal of the MCP Server is to give the connected LLM enough context to reason about platform behavior, not just report it. When a workflow surfaces a warning or error, the MCP Server can expose:

• Recent edit attribution on the affected pipeline or mapping group
• Audit history for related configuration changes
• Execution statistics from the run in question

With that context loaded, the LLM can provide insights and suggestions to help address processing warnings or errors, turning a generic notification into an informed troubleshooting conversation.

Open Standard, Open Architecture

The MCP Server preserves Nexadata's core architectural principle: raw customer data never leaves the platform. The MCP layer exposes metadata, state, and control to the LLM, while data processing remains entirely inside the deterministic Nexadata execution engine.

📄 View MCP Server documentation →

⏰ Runtime Variables for Parameter-Driven Workflows

Nexadata now supports Variables at the workflow level, enabling parameter-driven data processing across pipelines and mapping groups.

Built for Agent-to-Agent Automation

The headline use case for variables is runtime parameter passing in agentic automation flows. External agents and orchestration systems can inject parameter values into a Nexadata workflow at runtime, enabling dynamic, conditional data processing without modifying the underlying pipeline configuration.

This unlocks a class of workflows that previously required custom logic or duplicated pipeline definitions, including:

• Date-range scoped extractions where the range is decided by an upstream agent
• Multi-tenant processing, where the tenant identifier is passed in at run time
• Conditional transformations that depend on parameters resolved by an external system

How Variables Work

Variables are defined once at the workflow level and referenced throughout pipelines and transformations. They can be managed in two ways:

• In the Nexadata UI: A new Variable Management interface in the workflow lets users define, edit, and inspect variables directly
• Through the MCP Server: MCP-compatible clients can read and modify variable values through dedicated MCP tools, enabling LLM-driven parameter adjustment as part of a broader agentic workflow

📄 View Variables documentation →

📊 Execution Statistics on Workflow and Pipeline Runs

Nexadata now captures and surfaces per-run execution statistics for workflows and pipelines, giving users granular visibility into platform performance and execution behavior.

In the Platform UI

The Workflow Results view now shows execution statistics directly alongside each run, including duration, throughput, and stage-level metrics. This makes it straightforward to:

• Diagnose pipeline performance bottlenecks
• Compare run-to-run behavior across schedules
• Validate throughput characteristics during proof-of-concept evaluations

Same Statistics, Exposed via MCP

The execution statistics surfaced in the Nexadata UI are also exposed through MCP Server tools, so an MCP-compatible client draws on the same data backing the Workflow Results view. One source of truth, two consumption surfaces.

This is the connecting tissue between the platform UI and the MCP layer. A diagnostic conversation with an MCP client can reference live execution numbers identical to what a user sees on screen, ensuring consistent reasoning across human and agent workflows.

🧾 Enhanced Excel Support

This release significantly enhances Nexadata's support for Excel data sources, extending the platform's semi-structured data capabilities with native handling for spreadsheet-specific behaviors.

What's Enhanced

• Precise data region targeting: Define exactly which part of a workbook to load using Sheet Name, Anchor Cell, and Header Offset controls. Real-world spreadsheets with title rows, logos, banners, or gaps between the header and the first record can now be ingested cleanly without preprocessing.
• Dynamic Range detection: Auto-detect the contiguous data block starting from the Anchor Cell, adapting to datasets with varying row counts between runs. A fixed Rows-and-Columns mode is also available when the source file has a predictable shape.
• Flexible storage connections: Spreadsheet Datasets work with Nexadata's file-based connections, including Nexadata Hub, Amazon S3, and SFTP.
• Column fidelity: Source-file column ordering and multi-line header cells are preserved during ingestion.
• Smarter typing: Automatic column type detection respects numeric, date, and text content rather than defaulting everything to text.

📄 View Spreadsheet Dataset documentation →

🛠️ Platform Enhancements

This release includes a set of platform enhancements driven directly by customer feedback:

• ENG-935 – Improvement: Tag-based filtering is now available when adding datasets to a workflow
• ENG-926 – Improvement: An explicit "View Execution Details" button has been added to Workflow Results rows for better discoverability
• ENG-925 – Improvement: Timestamps across the platform now display in the user's local time instead of UTC
• ENG-955 – Improvement: A Delete button is now available directly on the Workflow Detail view
• ENG-939 – Improvement: Faster data onboarding performance for large Excel files
  ​

🧱 Platform Fixes

• ENG-440 – Bug Fix: Line feeds in column headers no longer disrupt CSV format processing
• ENG-914 – Bug Fix: Connector Copilot natural language search now correctly finds objects in large SAP C4C OData sources
• ENG-921 – Bug Fix: Copilot configuration logo updates now propagate to associated connections and datasets
• ENG-924 – Bug Fix: Calculated value columns from a supporting pipeline are now available in Group By aggregations in subsequent pipelines
• ENG-929 – Bug Fix: Deleting a supporting pipeline no longer breaks referential integrity in downstream pipelines that consume it
• ENG-936 – Bug Fix: Stack transformation is now available in the AI Copilot's tool set
• ENG-937 – Bug Fix: Transform Copilot now correctly resolves pipeline outputs as join sources in AI Mode
• ENG-945 – Bug Fix: Mapping Group selector in Build Mapping Transformation now supports search and alphabetical sort

🔭 What's Next

We're continuing to expand the agentic data integration footprint with a first-party companion to the MCP Server. Here's what we're working on:

• Nexadata Agents: A first-party, in-product chat surface with direct platform access, built natively into the Nexadata UI. No external client required.
• Page-aware context: Conversations open with the current page, workflow, or pipeline already in scope, so users can ask questions and take action without re-establishing context.
• Direct platform access: Native integration with the Nexadata execution layer for seamless reasoning and action across workflows, pipelines, and datasets.

Nexadata Agents will sit alongside the MCP Server within the broader agentic data integration story, providing users with a native in-product option that complements the external MCP client path.

Thanks for being part of the Nexadata journey!

The Nexadata Team