MirrorNeuron Developer Manual
Install the runtime, run blueprints, build workflows, and integrate your application.
MirrorNeuron runs packaged workflows called blueprints. Create a durable job, start a run, and inspect its logs and outputs with the CLI or HTTP API.
Start here
- Install MirrorNeuron and check runtime health.
- Run your first blueprint using reviewed sample inputs.
- Monitor the run, collect outputs, and handle human requests.
Choose your task
| I want to… | Guide |
|---|---|
| Find a blueprint and check its requirements | Choose a blueprint |
| Create a reusable workflow package | Author a blueprint |
| Use Python to load and compile a package | Python SDK |
| Select or diagnose a model | Model runtime |
| Start runs from my application | HTTP API |
| Automate CLI commands | CLI reference |
| Resolve a failed launch or missing output | Troubleshooting |
| Connect another machine | Connect runtime nodes |
| Review worker permissions and data access | Security |
Know these three terms
A blueprint is the reusable package. A job stores its configured definition. A run is one execution of that job. Use the job ID to start another execution, and the run ID to inspect, pause, resume, or cancel it. See Core concepts.
Begin with sample data and review each blueprint's worker code, model endpoints, and external actions before running it. A completed run still requires review of its outputs.