Enabling agentic AI: MCP server for AppsHub

The gap between AI assistants that talk about engineering work and AI agents that actually do it comes down to one thing: whether the agent can reach the tools. With the release of pSeven Enterprise v.2026.06, we are closing that gap for AppsHub. The new built-in MCP server exposes AppsHub directly to any MCP-compatible desktop AI agent, giving it a standardized interface to discover, inspect, and execute your simulation apps without custom integration work on your side.

In practice this means an AI agent working on your behalf can browse the apps available to you, read their input and output specifications, and run them – passing its own values, forwarding the output of one app as an input to another, uploading files, and tracking execution in real time. If something goes wrong, the agent can debug, interrupt, or clean up runs. What it cannot do is modify apps, create new ones, or access anything outside your existing permissions. The agent operates strictly within your access boundary – it sees what you see, nothing more.

This unlocks a qualitatively different class of automation. Scenarios that previously required manual orchestration, like running a parametric sweep across an app, feeding results into a downstream workflow, generating a report from the outputs – can now be driven by an AI agent through conversation or as part of a larger multi-agent system. The simulation expertise stays encoded in your AppsHub apps exactly as before; the MCP layer simply makes that expertise callable by an agent, without rebuilding or rewrapping anything.

Example of calling AppsHub simulation apps from OpenAI Codex via MCP

The MCP server ships as a built-in component of pSeven Enterprise v.2026.06 and requires no additional configuration to enable. If you are already running AppsHub workflows and using a desktop AI agent with MCP support, the two are now ready to talk to each other.