Prepare recurring analysis with AI automations

Before you start · Premium#

This guide covers tools included in Code Weaver Premium. Sign in to your account in the IDE and check that Premium is active. Without Premium you can still browse, search, edit and save your files.

Example: understand why a test failed#

After a failed test, you want a summary of affected files and useful checks. An automation keeps that instruction and prepares a new conversation when the test runner records another failure.

Create the instruction and try it#

1. Open Conversation tools → Automations and choose New automation.
2. Enter “Analyze failed tests” in Name. In Agent instruction, write: “Summarize the failure, identify affected files and suggest three checks in order of usefulness.”
3. Set Trigger to Failed tests. Leave Provider set to Project default or choose a connection and Model.
4. Choose Save automation, then Run now on the saved definition. Read the conversation that opens.
5. When the instruction is ready, choose Edit, check Enable automatic runs while Code Weaver is open and save again.

Choose when to prepare the next result#

Failed tests follows newly failed runs in the test runner. For a periodic check, choose At intervals and enter Interval in minutes. New local commit is useful for a changelog draft; Changed pull requests prepares analysis of project PR changes with GitHub configured.

For commits and pull requests, the first observation establishes the baseline. Subsequent changes trigger the saved instruction.

Result and next step#

Every run uses Plan and produces a result for review. Under Recent runs, choose Open result to read the analysis and continue the conversation. If you want a fix, describe the change in chat and start the normal proposal and review cycle.

To change the instruction or trigger, choose Edit. To pause future runs, uncheck automatic runs and choose Save automation; the result history remains available.