API/MCP¶
The Gradient Dynamics API/MCP service is the authoritative engineering interface. Every engineering operation goes through it, including operations initiated in Studio. REST and MCP expose the same validation, permissions, engineering requests and saved results. The Python client is a convenience wrapper around REST.
Studio provides the interactive design and simulation experience and manages project review and saved revisions. It uses the central service for engineering work. A returned candidate document or completed job does not, by itself, approve a Studio design change.
Pick an entry point¶
Entry point |
Use it for |
Start here |
|---|---|---|
Studio |
Visual design, setup, review and results |
|
REST |
Scripts, integrations and explicit HTTP control |
|
MCP |
Agent discovery and engineering tool calls |
|
Python client |
HTTP convenience methods and file streaming |
The service origin is https://api.gradient-dynamics.com; MCP is served at /mcp and REST engineering routes at /v1. Use the origin supplied for your environment. Feature availability is determined by that connected service, your permissions and the qualified capabilities it exposes.
A complete engineering workflow¶
Discover the required capabilities → import or create geometry → verify scale and regions → plan and validate the mesh → estimate and submit → inspect quality → configure and validate Fluxcore → estimate and submit → assess convergence and inspect saved results.
For machine reading, the built site publishes /llms.txt, /llms-full.txt and raw Markdown under /_sources. These aid discovery; live schemas still govern supported requests.
Keep the job and artifact IDs at every stage. Geometry, mesh and solution artifacts have different roles; do not substitute a display preview for a simulation mesh.
- Authentication and availability
- REST quickstart
- Agent workflow
- Projects, workflows and artifacts
- Uploads and downloads
- Units and coordinates
- Implicit geometry and CAD
- Design operation reference
- Meshing requests
- Fluxcore requests
- Thermal request example
- Rotating-region request example
- Turbulence and hybrid models
- Time integration and restart
- Advanced controls
- Spatial accuracy
- Linear convergence controls
- Output controls
- Jobs, operations and recovery
- Results and visualization
- Studies and predictions
- Errors and limits
- Python client
- MCP server
- REST endpoints and models
- Compatibility and migration