Handle errors and continue a run

Keep the failed step and its evidence in the same session. Choose the recovery action Prinevo shows, give the agent specific feedback when direction must change, and continue only after the required checks pass.

Keep the existing session. A retry preserves the failed execution as evidence and creates the next authorized attempt under the same workflow owner.

Current product guideVerified September 7, 2026

Read the failure before acting

Open the active session and find the stage marked Needs attention or Failed. Read the user-safe summary, the recommended next action, and any test, log, review, or validation evidence attached to the step.

Confirm that the failure belongs to the current stage, not an older attempt.
Open the latest evidence before asking the agent to change direction.
Keep the approved scope and design unless the failure proves they must change.

Choose the recovery path

What you see	What to do
Recovering workflow step	Prinevo is performing a bounded automatic retry. Wait for the attempt to finish instead of starting another retry.
Workflow step needs attention	Review the summary and evidence, then use the recommended retry action when the same step should run again.
Approval or review needs changes	Select Request changes or Type your own feedback, describe the correction, and submit it to the current stage.
Validation failed	Open the failed checks, ask the agent to fix the defect, and require those checks to pass before continuing.
Provider authentication required	Open Integrations, reconnect the provider, return to the same session, and use the available retry action.

Give corrective feedback

Use feedback when the agent needs new direction, not only another execution attempt. In a decision or review gate, select Request changes or Type your own feedback. You can also send the correction in the current session when the composer is available.

Simple recovery message

Can you fix this, rerun the failed checks, and move forward once they pass?

More specific recovery message

Fix the validation failure shown in the latest evidence. Keep the approved scope and design unchanged, rerun the affected checks, and continue to the next stage only after they pass. If the fix requires a scope or architecture change, stop and ask for approval.

Specific feedback is safer than “move ahead.” Name the failed behavior, the evidence to use, what must remain unchanged, and the checks that must pass.

Retry the failed step

1
Resolve required input first
Provide missing feedback, reconnect authentication, or address the blocker named by the session.
2
Select the retry action
Use Retry failed workflow step or the action recommended in the recovery card.
3
Wait for the current attempt
The action changes to Retrying. Do not click repeatedly or create a replacement session.
4
Review the new result
Prinevo keeps the earlier failure as history and displays the result of the authorized recovery attempt.

Confirm recovery before moving forward

A retry is successful only when the stage produces the required result and the affected checks pass. Review the updated artifact, code review outcome, and verification evidence before approving the next stage.

The original failure is addressed rather than hidden or bypassed.
Code Review has no unresolved blocking or actionable findings.
The failed tests or validation checks now pass.
The workflow advances to the expected next stage without creating duplicate work.