Abaqus Control MCP

中文 | English

Let AI drive Abaqus directly. Describe the model you want — geometry, materials, loads, steps — and let the AI work inside your live Abaqus/CAE session.

Abaqus Control MCP connects Claude Code, Codex, Antigravity, and any other MCP-compatible client to a running Abaqus/CAE instance. You describe the task, the AI translates it into Abaqus actions, and the model updates in real time.

Older Abaqus versions ship Python 2. If your Abaqus uses Python 2, use the Python 2 compatible fork instead.

Community Adoption and Ecosystem Impact

Since its open-source release, the designs and implementations of Abaqus-Control-MCP have been adopted and further developed by other open-source projects. We are glad to see this project advancing the integration of AI and engineering simulation and serving as a technical foundation for related tools.

• CAE-Agent-Hub adopted and adapted this project's live execution and diagnostic architecture in commit 5bd51ff on May 25, 2026, including the local socket bridge, GUI-thread dispatch, persistent execution context, and AST-driven error diagnosis and recovery mechanisms.
  • The relevant upstream implementation was available in the initial Abaqus-Control-MCP release, commit cc4ab0b, on May 10, 2026.
  • CAE-Agent-Hub provides attribution under the MIT License in its THIRD_PARTY_NOTICES.md.

We welcome and encourage developers to explore, improve, and build new tools upon Abaqus-Control-MCP. The value of open source lies in enabling good ideas to be reused, advanced, and shared with the broader community. If your project adopts designs or implementations from this project, feel free to open an Issue or submit a Pull Request to share your work and be included in this list. Together, we can explore more possibilities for AI-driven engineering simulation.

Why this?

• Work directly in the live GUI — actions run in your active Abaqus window, so geometry, mesh, and results stay visible as they change.
• Full access to Abaqus objects — mdb, session, odb, and the rest of the Python API are directly available, letting AI do the work.
• Keep the session interactive — engineers can inspect progress at any time without stopping the session.
• Stay local — the bridge listens on 127.0.0.1:48152, so nothing leaves your machine.
• Use any MCP client — Claude Code, Codex, Antigravity, and other compatible clients can all connect.

Setup

AI agent one-click install

If your AI agent supports natural-language installation, just prompt it with:

install https://github.com/Whfkl/Abaqus-Control-MCP

1. Install the package

You can install the package directly via Git, or from a cloned repository in your local directory.

Option A: Remote Git Install (direct install)

uv (recommended):

uv tool install git+https://github.com/Whfkl/Abaqus-Control-MCP.git

Or pip:

pip install git+https://github.com/Whfkl/Abaqus-Control-MCP.git

---

Option B: Local Directory Install (run inside the cloned project root)

• Standard Local Install

  uv (recommended):

  uv tool install .

  Or pip:

  pip install .

• Development Mode (editable install) If you plan to modify the source code and have changes apply dynamically:

  uv (recommended):

  uv tool install --editable .

  Or pip:

  pip install -e .

Dependencies are declared in pyproject.toml — no requirements.txt needed. Any of the methods above will install the four CLI commands (abaqus-control-mcp-server, abaqus-control-check, abaqus-control-doctor, abaqus-control-setup) to your global environment or active virtual environment.

2. Install the GUI plugin

Run the following command to install automatically:

abaqus-control-setup

Alternatively, you can manually copy abaqus_mcp_bridge/gui_plugin.py from the project source to your Abaqus plugins directory (typically ~/abaqus_plugins/).

Note

You can override the target folder for both automatic and manual installations by setting the ABAQUS_MCP_PLUGIN_DIR environment variable.

3. Start Abaqus/CAE, activate the plugin

Plug-ins → Abaqus-Control-MCP → Start MCP Bridge

4. Configure your MCP client

Depending on the AI agent you use, add the MCP server configuration to the corresponding config file.

Option A: Claude Code (Claude CLI)

Add the following entry to the mcpServers section in ~/.claude.json:

"abaqus-control-mcp": {
  "command": "abaqus-control-mcp-server",
  "env": {
    "ABAQUS_MCP_HOST": "127.0.0.1",
    "ABAQUS_MCP_PORT": "48152",
    "ABAQUS_MCP_TIMEOUT": "120"
  }
}

Note: Add to the mcpServers section in ~/.claude.json (global) or under projects.<path>.mcpServers for a specific project. Claude Code starts the MCP server automatically when you open a session.

Option B: Antigravity (Gemini/Antigravity IDE)

Add the following entry to the mcpServers section in ~/.gemini/config/mcp_config.json. Specifying an absolute path is recommended (replace <Username> with your system username):

"abaqus-control-mcp": {
  "command": "abaqus-control-mcp-server",
  "env": {
    "ABAQUS_MCP_HOST": "127.0.0.1",
    "ABAQUS_MCP_PORT": "48152",
    "ABAQUS_MCP_TIMEOUT": "120"
  }
}

5. Verify

abaqus-control-check

MCP Tools

Tool	Description
ping	Check connectivity + session state (models, viewports, PID)
run_python	Execute arbitrary Python code in the Abaqus kernel
monitor_job_status	List jobs or tail .sta/.msg for progress and diagnostics
inspect_odb	Open ODB read-only: sliced frames, variables with components
capture_viewport	Capture viewport screenshot as base64 (PNG/JPEG/TIFF/SVG)
set_workdir	Change the Abaqus working directory

Model creation, job submission, field/history output extraction — all go through run_python. Its error payload includes the traceback, error line, code excerpt, and recovery hints.

Environment Variables

Variable	Default	Description
ABAQUS_MCP_HOST	127.0.0.1	TCP bridge host
ABAQUS_MCP_PORT	48152	TCP bridge port
ABAQUS_MCP_TIMEOUT	120	Execution timeout in seconds
ABAQUS_MCP_PLUGIN_DIR	~/abaqus_plugins	GUI plugin target directory

Python API

from abaqus_mcp_bridge.client import AbaqusBridgeClient

client = AbaqusBridgeClient(timeout=60)
result = client.execute("from abaqus import mdb; result = list(mdb.models.keys())")
print(result['return_value'])  # ['Model-1', ...]

Troubleshooting

Issue	Solution
WinError 10061 Connection refused	The bridge service is not running in Abaqus/CAE. Please first start Abaqus/CAE, and select Plug-ins -> Abaqus-Control-MCP -> Start MCP Bridge from the top menu, then try again.
Permission denied / os error 5 during install or update	An old abaqus-control-mcp-server process is still running in the background (used and locked by Claude Code/Cursor). Close the MCP client (such as Claude Code or Cursor) or manually terminate the process in Task Manager to release the file lock, then try installing again.
MCP client says Connection closed / Server crashes with ModuleNotFoundError: No module named 'pydantic_core.core_schema'	The dependency packages (e.g. pydantic_core) in the default Python environment (e.g. Python 3.13) selected by the uv tool runner are corrupted or missing DLLs on Windows. Force a reinstall using another installed Python version: uv tool install --force --python 3.14 .
command not found: abaqus-control-mcp-server	Reinstall or run abaqus-control-doctor
No output from server	Normal for stdio MCP server — it doesn't print to stdout
Module abaqusGui can only be used in Abaqus/CAE GUI	Use Plug-ins menu, not File → Run Script
Connection timed out	Start the Abaqus plugin before the MCP server
Model doesn't appear in GUI	Run abaqus-control-check — verify "thread": "MainThread"
Claude Code doesn't see MCP tools	1. Run claude mcp list to check if abaqus is registered. 2. If not listed, add the MCP config as described in Step 4. 3. Restart Claude Code after any config change.

License

MIT — see LICENSE.