Skip to content

add checkpoint deployment cli support#721

Draft
kevinjosethomas wants to merge 3 commits into
mainfrom
improvement/checkpoint-deploy-cli
Draft

add checkpoint deployment cli support#721
kevinjosethomas wants to merge 3 commits into
mainfrom
improvement/checkpoint-deploy-cli

Conversation

@kevinjosethomas

@kevinjosethomas kevinjosethomas commented Jun 5, 2026

Copy link
Copy Markdown
Member
  • adds checkpoint deployment to the existing deployments create command with an explicit checkpoint id flag.
  • reuses the deployments api client so checkpoint deployment returns and displays the created adapter.
  • covers the endpoint call, success output, and conflict error behavior in deployment cli tests.

Note

Low Risk
Additive CLI/API client wiring with tests; no changes to auth or existing adapter deploy validation paths beyond a new optional flag.

Overview
Adds Hosted Training checkpoint deployment to the Prime CLI so users can start inference from a checkpoint without going through an existing adapter ID first.

DeploymentsClient gains deploy_checkpoint, which POSTs to /rft/checkpoints/{id}/deploy and parses the returned adapter like adapter deploy. prime deployments create accepts --checkpoint-id (mutually exclusive with positional MODEL_ID), runs confirmation unless --yes, prints adapter id/status, and reuses the same inference usage hints as model deploy.

The checkpoint path does not run the READY / deployable-base-model checks used for adapter deploy; it relies on the API. Tests cover the POST path, successful CLI output, and surfacing 409 conflict errors.

Reviewed by Cursor Bugbot for commit 13d3303. Bugbot is set up for automated code reviews on this repo. Configure here.

@cursor cursor Bot left a comment

Copy link
Copy Markdown

Choose a reason for hiding this comment

The reason will be displayed to describe this comment to others. Learn more.

Cursor Bugbot has reviewed your changes and found 1 potential issue.

Fix All in Cursor

❌ Bugbot Autofix is OFF. To automatically fix reported issues with cloud agents, enable autofix in the Cursor dashboard.

Reviewed by Cursor Bugbot for commit 13d3303. Configure here.

_print_inference_usage(adapter.base_model, adapter.id)
return

assert model_id is not None

Copy link
Copy Markdown

Choose a reason for hiding this comment

The reason will be displayed to describe this comment to others. Learn more.

Empty checkpoint-id triggers assert

Medium Severity

Passing --checkpoint-id with an empty value skips the checkpoint deploy path and the no-argument help exit, then hits assert model_id is not None. That assertion is not caught as APIError, so the CLI can crash instead of showing a clear validation error.

Additional Locations (1)
Fix in Cursor Fix in Web

Reviewed by Cursor Bugbot for commit 13d3303. Configure here.

@kevinjosethomas kevinjosethomas marked this pull request as draft June 5, 2026 02:55
Sign up for free to join this conversation on GitHub. Already have an account? Sign in to comment

Labels

None yet

Projects

None yet

Development

Successfully merging this pull request may close these issues.

1 participant