Choosing an interface

opentide exposes the same engine three ways. They share the object model, validation, and deployment logic — the difference is who is driving.

flowchart TD
  engine["opentide engine<br/>objects · validate · generate · deploy · document"]
  cli["CLI · opentide"] --> engine
  sdk["SDK · from opentide import Opentide"] --> engine
  mcp["MCP · opentide mcp start"] --> engine

Decision table

Rules of thumb

• Start with the CLI. For onboarding, daily authoring, and CI, the CLI is the shortest path. Everything in the tutorial uses it.
• Reach for the SDK when you need to compose. Building a custom report, a bespoke gate, or integrating opentide into another tool? The SDK gives you the objects and results as Python.
• Use MCP for agents, not automation. MCP is for interactive AI assistance in an editor. For unattended pipelines, use the CLI — and remember MCP's validate_query and run_query contact a tenant only with live: true (agentic setup).

The same operation, three ways

Validating a rule:

opentide validate --uuid 00000000-0000-4000-8003-000000000001 --strict

from opentide import Opentide
Opentide.initialise()
result = Opentide.Rules["00000000-0000-4000-8003-000000000001"].validate()
print(result.ok, result.errors)

validation_report(uuid="00000000-0000-4000-8003-000000000001")

Capabilities compared

Next

• CLI reference
• SDK reference
• MCP reference