Write Your First Operator

Audience: extension authors
Status: Preview

The current external operator path uses a manifest plus a Python entrypoint.

1. Create the Python entrypoint

from ophiolite_sdk.operators import OperatorRegistry, computed_curve

registry = OperatorRegistry()

@registry.register("acme:double")
def double_curve(request):
    curve = request.payload["inputs"]["input_curve"]
    values = [None if value is None else value * 2.0 for value in curve["values"]]
    return computed_curve("DTX", values, unit=curve.get("unit"))

2. Declare the operator package

{
  "schema_version": 1,
  "package_name": "acme-tools",
  "package_version": "0.1.0",
  "provider": "acme",
  "runtime": "python",
  "compatibility": { "ophiolite_api": "0.1.0" },
  "entrypoint": "acme_ops.py",
  "operators": [
    {
      "id": "acme:double",
      "provider": "acme",
      "name": "Double Curve",
      "asset_family": "Log",
      "category": "Demo",
      "description": "Doubles a selected curve.",
      "default_output_mnemonic": "DTX",
      "output_curve_type": "Computed",
      "input_specs": [
        {
          "SingleCurve": {
            "parameter_name": "input_curve",
            "allowed_types": []
          }
        }
      ],
      "parameters": [
        {
          "Number": {
            "name": "factor",
            "label": "Factor",
            "description": "Scale factor",
            "default": 2.0,
            "min": 0.0,
            "max": 10.0,
            "unit": null
          }
        }
      ],
      "output_lifecycle": "derived_asset",
      "deterministic": true,
      "stability": "preview",
      "tags": ["demo"]
    }
  ]
}

3. What Ophiolite still owns

Even for external operators, Rust still owns:

• project locking
• input and parameter validation
• execution contracts
• provenance
• derived-asset persistence

Next: Package and install an operator