About loadbench
loadbench is an MCP server that gives AI assistants structural load and stability calculations, including tipping, support reactions, and beam stress and deflection checks. Its calculations are verified against hand-computed answers and are intended for planning estimates, not certified engineering.
Key features
- 01Check whether loads cause tipping
- 02Calculate leg or bracket reactions
- 03Check beam stress and deflection
- 04Verify calculations against hand-worked results
Use cases
- 01Assess whether an object may tip under load
- 02Estimate forces on legs or brackets
- 03Make an initial load check for a shelf or beam
MCP support
The server exposes the check_tipping, solve_supports, and beam_check tools to MCP clients for structural load and stability calculations.
Configuration
Local server config is the same for every client. Copy it into your client’s MCP config file.
Tools
3Saved tool list · updated 2w ago
| Parameter | Type | Required | Description |
|---|---|---|---|
| masses | array | Yes | |
| base_polygon | array | Yes |
Details
From the project READMEloadbench-mcp
An MCP server that does the structural load & stability math language models get wrong. AI assistants call it instead of guessing.
| Tool | Question it answers |
|---|---|
check_tipping | Will these weights topple over their footprint? |
solve_supports | How much load does each leg / bracket carry? |
beam_check | Will this beam or shelf hold the load? (stress + deflection) |
All math is verified against hand-computed answers (tests/test_physics.py).
Install & run
Once published to PyPI, any MCP client can launch it with no install step:
uvx loadbench-mcp
Or install it:
pip install loadbench-mcp
loadbench-mcp
Use it in Claude Desktop
Add to your claude_desktop_config.json:
{
"mcpServers": {
"loadbench": {
"command": "uvx",
"args": ["loadbench-mcp"]
}
}
}
Restart Claude Desktop, then ask things like "A 2 m pine shelf on two brackets 0.3 m from each end — will it hold 40 kg in the middle?"
Local development
uv run --with mcp --with numpy python -m load_solver.server # run from source
python -m pytest tests # verify the math
Inputs are SI units (metres, kilograms, newtons; 1 kg ≈ 9.81 N). Tool docstrings carry the full argument details, which the model reads automatically.
Estimates for planning, not certified engineering. For loads where a failure could cause injury or real damage, have a qualified engineer check the result.
License
MIT
Comments
New to MCP?
From what the protocol is to installing your first server, these answers cover most first questions.
An open protocol that defines how AI applications connect to external tools and data in a uniform way. Clients talk to the model; MCP servers provide tools, resources and prompt templates. They exchange standard messages, so one server can be reused by many clients.