Skip to main content

AI Runner

The AI Runner performs AI-driven analysis on repository code changes (diffs) associated with a specific build. It evaluates what has changed between revisions to assess potential build risk and generate the AI Build Analysis ReportAI Build Analysis Report is a document-style report integrated into the Build Results in AI Analysis tab. It provides automated build risk assessment along with detailed file-level and code-level insights for completed b….

The AI Runner executes its analysis asynchronously alongside the build workflow, ensuring that AI processing does not block or interfere with core CI/CD execution.

When to Use​

Use the AI Runner when you need to:

  • Analyze code changes introduced in a build.
  • Assess build risk based on recent commits or pull request diffs.
  • Generate an AI Build Analysis Report focused on change impact.
  • Add AI-based insight without modifying build execution logic.

Parameters​

ParameterDescription
Step NameOptional step name. Defaults to the runner name if not specified.
Working DirectoryDirectory in which the AI analysis runs. Specify only if different from the repository checkout directory.
App KeyApplication key used to authenticate with the AI service.
App SecretApplication secret used for secure authentication.
Model TierSelects the level of AI analysis: Economy, Optimized, or Best.
Failure Percentage ThresholdBuild failure probability (0–100) above which the step fails the build.
Continue on Server Error

Specifies how the build should behave if the AI analysis fails due to a server-side issue (for example, KenzoAI service unavailability or analysis errors).

  • Fail on server error: Marks the build step as failed if a server error occurs during AI analysis.
  • Ignore server error: Skips the AI analysis failure and allows the build to continue with subsequent steps.

ai-runner

Typical Use Cases​

Common scenarios for using the AI Runner include:

  • Evaluating pull request changes before merge.
  • Identifying risky code modifications early in the pipeline.
  • Adding change-impact awareness to CI workflows.
  • Flagging builds likely to fail based on recent code changes.

Example​

This example demonstrates how to add an AI Runner to a build configuration.

Scenario
Analyze code changes for every commit on the main branch.

Configuration settings:

  • Step Name: AI Code Diff Analysis
  • Working Directory: (default)
  • App Key: <configured-app-key>
  • App Secret: <configured-app-secret>
  • Model Tier: Optimized
  • Failure Percentage Threshold: 65
  • Continue on Server Error: Fail on server error

Limitations​

  • The AI Runner analyzes only code diffs.
  • Requires valid KenzoAI credentials configured by an administrator.
  • AI insights are advisory and may not cover highly customized or non-standard workflows.
  • Higher model tiers may increase analysis time.

Best Practices​

  • Place the AI Runner early in the pipeline to detect issues before downstream steps.
  • Select the model tier based on required depth and build criticality.
  • Review AI findings alongside build context and human judgment.
  • Monitor analysis results over time to identify recurring risks or patterns.
  • Use failure thresholds carefully to avoid blocking pipelines unnecessarily.