Setup & helpers
handoff
Delegate a workflow step to the CodeLantern cloud agent to run on its own.
Setup & helpers
What it does
/cl-dev:handoff delegates a step of your workflow to the CodeLantern cloud agent, which picks it up and runs it autonomously. Instead of the skill running interactively in your session, the cloud agent carries out the work against the same issue or pull request and reports back through it. You can step away while a plan gets drafted or a change gets built.
When to use it
Reach for it when a step doesn't need you sitting in front of it, most often planning or implementation, once the direction is clear. It's the same workflow, running somewhere else:
… claim-issue → create-impl-plan → approve-plan → implement → …Good moments to hand off:
- A plan is approved and you want
/cl-dev:implementto build it while you focus on something else. - You want the cloud agent to claim a spec'd issue and draft its implementation plan so it's waiting for you to review.
- You'd rather run a longer step unattended than keep a local session open through it.
Keep a step local when you expect to steer it turn by turn, especially interviews and /cl-dev:design-solution.
Prerequisites
The cloud agent must be set up for the repository first. It runs on GitHub repositories only. Run handoff from your own local session, signed in to CodeLantern with your personal account. Organization API keys can't post cloud commands. See Cloud agent for its requirements, the other commands you can post yourself, and where to find a run's results.
How to invoke it
From within your coding agent, in the repository you're working in:
/cl-dev:handoffIt posts a @codelantern-ai /architect comment on the issue to hand off planning, or a @codelantern-ai /implement comment on the pull request (or the issue, if there's no pull request yet) to hand off the build, and shows you the comment before posting it. Make sure the step is ready: /architect needs a spec'd issue, and you'll usually approve the plan before you hand off the build. If you're returning to work in progress, run /cl-dev:set-context first to bring the current state back into your session, then hand off from there.
Setup & helpers
Related skills
Other skills in the same phase of the workflow.
| backlog | Review and prioritize your open issues before you commit to the next piece of work. |
| cl-init | Initialize a repository so CodeLantern workflows can run against it. |
| doctor | Run a health check of your CodeLantern setup and get told exactly what to fix. |
| set-context | Load the state of an in-progress work item so you can pick up where you left off. |
| size | Assess an issue's complexity on an XS–XL scale and record it on the issue. |