Clone a run - Cloud and Local - SDK and CLI

[sebastian-quintero]

Description

Introduces functionality for cloning a run in cloud and local:

• SDK: new methods for interacting with the cloud module ➡️ cloud.Application.clone_run and cloud.Application.clone_run_with_result.
• SDK: new methods for interacting with the localmodule ➡️ local.Application.clone_run and local.Application.clone_run_with_result.
• CLI: new commands for cloning runs ➡️ nextmv cloud run clone and nextmv local run clone.

Deprecates some features. Largely, when run "metadata" is requested, it always comes back to the user as:

{
  "description": "",
  "id": "latest-6O8pEx1vg",
  "metadata": {
    "application_id": "2005",
    "application_instance_id": "latest",
    ...
  },
  "name": "",
  ...
}

It has always been confusing to get metadata and find metadata to be a nested field that is part of that object. In the SDK, the complete object has always been called RunInformation. Following this, I am proposing we move away from metadata methods in favor of information methods. These are the following features that are being deprecated:

• SDK: cloud.Application.run_metadata in favor of cloud.Application.run_information.
• SDK: local.Application.run_metadata in favor of local.Application.run_information.
• CLI: nextmv cloud run metadata in favor of nextmv cloud run information.
• CLI: nextmv local run metadata in favor of nextmv local run information.

Last but not least, missing information in the actual run metadata was completed, such as options, execution_class, and others...

[Copilot]

Pull request overview

This PR adds “clone run” support across the SDK (cloud + local) and CLI, and standardizes the “run metadata” concept by introducing run_information APIs/commands while deprecating run_metadata. It also expands the run information schema to include additional runtime fields (options, queuing, integration, execution details, etc.).

Changes:

• SDK: add clone_run / clone_run_with_result for cloud and local applications; add run_information and deprecate run_metadata.
• CLI: add nextmv cloud run clone + nextmv local run clone, and add run information commands while deprecating run metadata.
• Model/schema: extend nextmv.run.Metadata (options, queuing, integration, execution fields) and add new option/integration types.

Reviewed changes

Copilot reviewed 27 out of 27 changed files in this pull request and generated 14 comments.

Show a summary per file

File	Description
nextmv/tests/local/test_application.py	Updates local tests to use run_information and patches accordingly.
nextmv/tests/integration/cloud/test_integration_cloud.py	Adds an integration step covering cloud run cloning flows.
nextmv/nextmv/run.py	Extends run models (options/integration/etc.) and adds content_format helper + new types.
nextmv/nextmv/local/runner.py	Records richer local run metadata (options, tracking, queuing, format).
nextmv/nextmv/local/executor.py	Updates local run info to include execution_duration.
nextmv/nextmv/local/application.py	Adds local clone_run*, introduces run_information, and deprecates run_metadata.
nextmv/nextmv/input.py	Adjusts option resolution timing during Input.load.
nextmv/nextmv/deprecated.py	Introduces NextmvDeprecationWarning and updates warning emission behavior.
nextmv/nextmv/cloud/application/_run.py	Adds cloud clone_run*, introduces run_information, refactors run submission into __new_run, deprecates run_metadata.
nextmv/nextmv/cli/mcp/tools/run.py	Switches MCP “cloud run status” to run_information and minor formatting changes.
nextmv/nextmv/cli/mcp/tools/local.py	Switches MCP local run status/result validation to run_information and minor formatting changes.
nextmv/nextmv/cli/main.py	Suppresses SDK deprecation warnings for CLI execution.
nextmv/nextmv/cli/local/run/metadata.py	Deprecates local CLI run metadata and prints a deprecation warning.
nextmv/nextmv/cli/local/run/input.py	Uses run_information to determine input format behavior.
nextmv/nextmv/cli/local/run/information.py	Adds nextmv local run information command.
nextmv/nextmv/cli/local/run/get.py	Uses run_information and avoids trimming empty assets.
nextmv/nextmv/cli/local/run/create.py	Allows “no input” in shared helper (for cloning) and updates examples.
nextmv/nextmv/cli/local/run/clone.py	Adds nextmv local run clone command.
nextmv/nextmv/cli/local/run/init.py	Registers new clone and information subcommands.
nextmv/nextmv/cli/cloud/run/metadata.py	Deprecates cloud CLI run metadata and prints a deprecation warning.
nextmv/nextmv/cli/cloud/run/input.py	Uses run_information to determine input format behavior.
nextmv/nextmv/cli/cloud/run/information.py	Adds nextmv cloud run information command.
nextmv/nextmv/cli/cloud/run/get.py	Uses run_information and avoids trimming empty assets.
nextmv/nextmv/cli/cloud/run/create.py	Allows “no input” in shared helper (for cloning).
nextmv/nextmv/cli/cloud/run/clone.py	Adds nextmv cloud run clone command and run-configuration overrides.
nextmv/nextmv/cli/cloud/run/init.py	Registers new clone and information subcommands.
nextmv/nextmv/init.py	Exposes new run model types at package top-level.

Comments suppressed due to low confidence (2)

nextmv/nextmv/cli/cloud/run/create.py:526

• resolve_input_kwarg now checks if stdin: (truthiness) instead of if stdin is not None. If the user pipes an empty input (after .strip()), this will be treated as “no stdin”, potentially returning {} and leading to confusing downstream behavior.

    # It is possible to not provide any input when we are cloning a run.
    if stdin is None and input is None and managed_input_id is None:
        return {}

    if stdin:
        # Handle the case where stdin is provided as JSON for a JSON app.
        try:
            input_data = json.loads(stdin)
        except json.JSONDecodeError:
            input_data = stdin

        return {"input": input_data}

nextmv/nextmv/cli/cloud/run/create.py:526

• resolve_input_kwarg now checks if stdin: (truthiness) instead of if stdin is not None. If the user pipes an empty input (after .strip()), this will be treated as “no stdin”, potentially returning {} and leading to confusing downstream behavior.

    # It is possible to not provide any input when we are cloning a run.
    if stdin is None and input is None and managed_input_id is None:
        return {}

    if stdin:
        # Handle the case where stdin is provided as JSON for a JSON app.
        try:
            input_data = json.loads(stdin)
        except json.JSONDecodeError:
            input_data = stdin

        return {"input": input_data}

---

💡 Add Copilot custom instructions for smarter, more guided reviews. Learn how to get started.

[merschformann]

Looks good, no major comments. Feel free to ignore them, so, already 🟢 it. Thanks for the work! 🤗

Regarding the metadata -> information rename, I am fine with that, though, it does drift from the API endpoint name (...runs/{run_id}/metadata). However, it is odd that it has nested metadata and for a CLI user it's likely nicer.
We need to remember to align nextpipe too though. 😊

[merschformann]

@sebastian-quintero : We need to check why smoketest is suddenly failing on Windows. Ping me if you need help around testing / finding the issue on Windows. ☺️

[sebastian-quintero]

Looks good, no major comments. Feel free to ignore them, so, already 🟢 it. Thanks for the work! 🤗

Regarding the metadata -> information rename, I am fine with that, though, it does drift from the API endpoint name (...runs/{run_id}/metadata). However, it is odd that it has nested metadata and for a CLI user it's likely nicer. We need to remember to align nextpipe too though. 😊

I think it is fine to drift from the API as long as it proves useful and makes more sense. I always found it confusing that some information is under metadata and other info is at the "root", like name... At least now we have a very clear distinction between information and metadata.