Skip to Content

Runs

Start a development run, follow its progress and logs, stop or retry it, and find the pull requests it opened.

A run is CodeLynq doing the work of a task: it checks out the repository, makes the change described in the approved spec, runs your checks, reviews its own work and opens a pull request. A task can have several runs - for example a retry after a failure, or a new run with your feedback.

Start a run

A run can start once the task's spec is approved. Open the task, go to the Runs tab and click Trigger New Run. Next to the button you may see an estimate of what a run costs, based on earlier runs in this project.

You can also:

Any project member can start a run. A run does not start when the organization has no active plan, has reached its plan's limits or has used up its model credit; you see why right away. See Billing and credits. If your organization has set spend limits, an expensive run may wait for approval first - see Approving expensive runs.

Statuses

StatusMeaning
Needs approvalHeld for its estimated cost until someone approves it. See Approving expensive runs.
QueuedWaiting to start.
RunningWorking.
CompletedFinished; see its pull requests.
FailedStopped with an error; see why below.
CancelledStopped by someone, or not approved.

Runs executed on your own machines carry a Self-hosted runner badge. See Self-hosted runners.

Follow a run

Select a run on the Runs tab to see:

  • Execution Logs - what the run is doing, live while it runs (Streaming...).
  • Changed Files - each file created (+) or modified (~).
  • Details: Branch, Commit, Started, Completed and the Model used.

Runs in the sidebar lists the runs of all projects, with task, project, status, time, cost and a link to the pull request.

Results and pull requests

When a run completes, it has opened a pull request in your git host:

  • With one repository: click View Pull Request # followed by its number.
  • With several repositories or branches: the Repositories list shows each repository and branch, marked Changed or No changes, with a link to each pull request. A repository or branch without changes gets no pull request.

What was checked lists each quality gate of the run, per repository: your verify command, the self-review, the security scan, the architecture rules, the docs step and CI on the pull request, each Passed, Failed, Not run, Switched off, After the PR or Waiting, with the command and the reason where something did not run. A finished run means the work is done and a pull request is open; only what is marked Passed was checked. The full output is under Technical logs.

Before you start a run, the Runs tab says whether it can start: an approved spec, a repository, your plan and model credit, free capacity, a runner (self-hosted) and your spend limits, with the expected cost and a link to fix what is missing.

Under Pull request feedback you see whether the pull request is mergeable or has conflicts, the CI checks and the Review comments. Click Refresh to update them. Review and merge in your git host as usual. If CI fails or a reviewer asks for changes, see The CI feedback loop.

When a run fails

A failed run explains in plain language what went wrong and what to do next. When it says This one is worth simply trying again., a retry is likely to succeed. Technical detail shows the original error.

Common reasons include a verify command that kept failing (see Quality gates), and repository access (see Connect a repository).

Stop, requeue and retry

  • Stop run - stops a queued or running run. It ends as Cancelled.
  • Requeue - puts a queued run back in the queue.
  • Retry run - starts a failed run again. A retry may use a stronger model; the run's Model detail then says which model it was escalated from.

Feedback for the next run

On the Feedback sub-tab of Runs you can collect comments on the work:

  1. Under Add Feedback, type your feedback, choose General Comment or Development Feedback, and click Add.
  2. Tick the development feedback items you want addressed.
  3. Click Trigger Run with Feedback. The new run gets your feedback along with the spec.

Once the task has an open pull request, the button reads Work into the spec as a change instead. See Request a change: the change is built on the same branch and no second pull request is opened.

To have CodeLynq address CI results and review comments on an existing pull request instead, use Iterate on feedback - see The CI feedback loop.

Notifications

To get a push notification when a run completes or fails, turn on Run completed or Run failed under Security settings → Notifications, and install the mobile app. Your organization can also send run notifications to Slack or a webhook - see Integrations.

Test it on a task from your own backlog

Send us one real task. We run CodeLynq on your repository and you review the pull request. Free, within 5 working days of approving the spec.