Skip to content

Microbots Log Analyzer for Azure Pipelines

A follow-along guide for adding Microbots' LogAnalysisBot to an Azure Pipeline as a custom task. By the end, failing builds will get an automatic root-cause analysis printed straight into the Pipeline.

Who is this for? Anyone running an Azure DevOps pipeline who wants automated analysis of build or runtime logs. The task authenticates to Azure OpenAI through an Azure Resource Manager Service Connection, creates an isolated Python venv on the agent, installs microbots[azure_ad], and reports the result.

Prerequisites

  • Azure DevOps organization where you can install custom extensions.
  • Azure Resource Manager Service Connection with permission to request tokens for the Azure OpenAI resource. See Azure managed identity setup for service connection and Azure OpenAI RBAC setup. The pipeline must be authorized to use this service connection.
  • Azure Pipelines agent with azure-cli, python3, pip and python3 -m venv support. Microbots uses Docker sandboxing by default, so the agent also needs a reachable Docker-compatible daemon.
  • Azure OpenAI deployment that works with Microbots and is reachable by the service connection.

Step 1 — Install the Marketplace Extension in Azure DevOps (Pipelines)

Install the published Marketplace extension into the Azure DevOps organization that owns your pipelines:

  1. Open the Marketplace listing: Microbots Log Analyzer. You may be required to sign in to Marketplace to view the extension.
  2. Select Get it free.
  3. Choose the Azure DevOps organization where your pipelines run.
  4. Confirm the extension appears in the organization extension page: https://dev.azure.com/<organization>/_settings/extensions. If it appears under the Shared section you need to install it to your org (requires Azure DevOps organization administrator access).

Step 2 — Add the Task to a Pipeline YAML

See the complete sample pipeline at microbots-log-analyzer.yml.

- task: MicrobotsLogAnalyzer@0
  displayName: Analyze build log
  inputs:
    azureSubscription: my-azure-service-connection
    deploymentName: my-azure-openai-deployment
    endpoint: https://my-azure-openai-resource.openai.azure.com/
    apiVersion: 2025-03-01-preview
    codebasePath: $(Build.SourcesDirectory)
    logFilePath: logs/build.log
    outputFilePath: $(Build.ArtifactStagingDirectory)/microbots-log-analysis.md
    additionalContext: |
      This build usually fails when package version conflicts occur.
      Please consider it while analyzing the log.
    timeoutSeconds: 600
    maxIterations: 20

The log file must exist before MicrobotsLogAnalyzer@0 runs. Relative logFilePath values are resolved from codebasePath; absolute paths are also supported.

outputFilePath is optional. When it is provided, it must be an absolute path ending in .txt, .md, or .log. The file does not need to exist; the task creates missing directories and replaces any existing file content with the latest LLM analysis result.

additionalContext is optional. When provided, it is appended as extra user context for the log analysis and does not replace or override the Microbots system prompt. Maximum length: 1024 characters.

✅ That's it! Your pipeline will now automatically analyze the provided logs and report the root cause.

📑 Reference — the sections below are reference material for tuning and understanding the task.

Inputs

Input Required Default Description
azureSubscription Yes - Azure Resource Manager service connection used for Azure CLI login. Alias: serviceConnection.
deploymentName Yes - Azure OpenAI deployment name.
endpoint Yes - Azure OpenAI endpoint, for example https://your-resource-name.openai.azure.com/.
apiVersion Yes - Azure OpenAI API version passed to Microbots, for example 2025-03-01-preview.
codebasePath Yes - Repository or source folder Microbots can inspect while analyzing the log.
logFilePath Yes - Log file path. Use an absolute path, or a relative path resolved from codebasePath.
outputFilePath No - Absolute .txt, .md, or .log path where the LLM analysis result is written. Missing directories are created, and existing file contents are replaced.
additionalContext No - Additional user context appended to the log analysis prompt. Maximum length: 1024 characters.
timeoutSeconds No 600 Maximum time for LogAnalysisBot.run().
maxIterations No 20 Maximum number of Microbots iterations. Leave unset to use the default from LogAnalysisBot.run().

How It Works

  1. Azure Pipelines runs the task with the Node20_1 task handler.
  2. The task logs in with the supplied Azure Resource Manager Service Connection.
  3. The task creates or reuses a virtual environment (microbots-log-analyzer-venv).
  4. The task installs microbots[azure_ad] into that virtual environment.
  5. A short Python runner creates LogAnalysisBot with AzureCliCredential, mounts codebasePath as context, passes logFilePath, optional additionalContext, optional maxIterations, and timeoutSeconds to LogAnalysisBot.run(), and prints the analysis result.
  6. If outputFilePath is provided, the task writes the LLM analysis result to that file, replacing any existing contents.

The task clears the Azure CLI account at the end of the run. Its task manifest also uses Azure Pipelines command restrictions so analyzed log content cannot run arbitrary logging commands or set pipeline variables.

Next Steps