Skip to content
Tresca

NeqSim MCP servers and AI agent skills

Equinor, which develops NeqSim, publishes an official MCP server that runs flashes, phase envelopes and JSON-defined process models, plus about 130 core skills in the NeqSim repository and a separate catalogue of 60 community screening skills. Everything is Apache-2.0 and runs on Windows, macOS or Linux, with the server needing Java 21 or Docker. The skills are written mainly for VS Code Copilot, with exports for other agents.

  • Includes Equinor (open source)'s official server.
  • 1 MCP server and 2 agent skills listed.
Equinor (open source)1 server · 2 skillsChecked 4 Nov 2026

Integrations for NeqSim

Official

SkillOfficial

NeqSim core skills and agentsby Equinor

About 130 SKILL.md files shipped in the NeqSim repo, from flash convergence to field development.

Windows, macOS, Linux163· updated 1 day ago
MCP serverOfficial

NeqSim MCP Serverby Equinor

Equinor's own MCP server for NeqSim: flashes, phase envelopes and JSON-defined process runs.

LocalWindows, macOS, LinuxRead-only or read-write (configurable)163· updated 1 day ago
SkillOfficial

NeqSim Community Skillsby Equinor

60 public screening skills for NeqSim in process, PVT, flow assurance, safety and subsea work.

4· updated 4 days ago

Listings are sorted by origin, then by recent activity. GitHub data is refreshed nightly. Nothing here is ranked.

How AI connects to NeqSim

An MCP server or skill can only do what NeqSim lets programs do. NeqSim exposes 2 automation interfaces:

  • Java library (Java): NeqSim is a Java library for fluid behaviour, phase equilibrium and process simulation, published to Maven Central. You define a fluid with an equation of state, add components, run flashes, then connect streams, separators, compressors and other equipment into a process system and run it. The MCP server is a Quarkus application built on this library, so every tool call is a Java calculation, not a lookup. Documentation
  • Python package (neqsim) (Python): pip install neqsim gives Python access to the same Java engine, which suits notebooks and the task-solving agents. Many of the community skills ship small Python packages that call it, and fall back to placeholder logic if the package is missing, so confirm it is installed before you rely on a skill's numbers. Documentation

What Equinor (open source) offers

Equinor runs NeqSim as an agentic engineering project: the main repository contains the MCP server with three tool tiers and four deployment profiles, core skills and agents under .github, a neqsim CLI to install community and private skill catalogues, and a task solver that writes Word and HTML reports. These are first-party releases, not a separate product.

Checked 4 Nov 2026. Sources: NeqSim README, "Agentic Engineering & MCP Server", NeqSim Skills and Agents Guide

What you can do with NeqSim and AI

  • Ask for a dew point, hydrate temperature or density at given conditions and get a NeqSim flash result with the equation of state and convergence status attached.
  • Draw a phase envelope for a natural gas composition, or a property table across a pressure range.
  • Simulate a short process such as separation and compression to export pressure from a JSON definition, and read back stream conditions and compressor power.
  • Screen TEG dehydration, hydrate margin, produced-water scale or vacuum collapse with the community skills, then hand the result to an engineer for review.
  • Run the server in an advisory-only profile for operators, so tools that change sessions or write variables are blocked.

Taken from the listings' own documentation. Check each listing for exactly what it supports.

Installing

The listings here document setup for Claude, Claude Desktop, Codex, Cursor and VS Code (GitHub Copilot). Install steps, copied from each project's own documentation, are on each listing's page. Most NeqSim integrations run on your own computer next to NeqSim, so you need NeqSim installed and licensed on the same machine unless the listing says otherwise.

Questions

Is there an official NeqSim MCP server?

Yes. Equinor publishes it in the neqsim repository, with a runner jar attached to each release and a Docker image on GitHub's container registry. Tresca marks it official pending the site owner's review.

Do I need a licence to use NeqSim with Claude?

No. NeqSim and its MCP server are Apache-2.0. You need Java 21 or later for the jar, or Docker, and an MCP client such as Claude Desktop, VS Code or Cursor.

How are the core skills different from the community skills?

Core skills live in the NeqSim repository and change with its API, covering topics from flash convergence to field development. Community skills sit in a separate repository, ship small Python packages with tests, and are described by Equinor as screening and learning aids, not design tools.

Can the server be used against live plant data?

It has a DIGITAL_TWIN profile that is advisory only: it computes answers but can't write back to operational systems, and state-changing tools are blocked. Connecting it to plant data is up to your own integration.

AI-generated models, drawings and results must be checked by a qualified engineer before use.

Sources (checked 4 Nov 2026): NeqSim documentation · equinor/neqsim README · NeqSim MCP Server README · equinor/neqsim-python · NeqSim product page

NeqSim is developed by its own project (Equinor (open source)). Tresca is independent and not affiliated with or endorsed by the NeqSim project.