Home
An AI-agent-native toolkit for fatigue analysis of materials: a Python library plus an MCP server with 44 tools, so AI agents can run the whole analysis by calling tools. General purpose and material agnostic. All analysis uses true stress and true strain, and the fatigue exponents b and c are negative throughout.
By David Fieser and Hugh Shortt. Both authors contributed equally.
- Getting Started: install the library, the MCP server, the no-code GUI, or the standalone Windows app, and run a first analysis.
- MCP Client Setup: register the server with Claude Desktop, Claude Code, Cursor, VS Code, or Google Antigravity.
- Agent Usage: the MCP tools, their units, and the compute, save, recall pattern for AI agents driving the toolkit.
- Physics Review: every equation implemented in the toolkit, defined and cited, written for review by a materials or fatigue specialist.
- Validation: what is validated against which published values, with the honest results, including where predictions run non-conservative.
- FAQ: common questions, answered plainly.
- Troubleshooting: known rough edges and their fixes.
- README: overview, install, and quick start.
- Documentation site: installation, usage, the SAE 1137 tutorial, the statistics guide, and the API reference.
- Landing page and setup guide
- Changelog
- Typeset physics PDF with reviewer sign-off table
- Cite the software with the "Cite this repository" button on the repo page, or through the Zenodo concept DOI 10.5281/zenodo.21222820.
The Agent Usage and Physics Review pages mirror files in the repository. The canonical copies live in docs/, and the wiki is updated when they change.