Azure DevOps Marketplace Extension

Code reviews that
actually ship

Claude reviews every pull request on GitHub, GitLab, and Bitbucket — all triggered from your Azure DevOps pipeline. No servers to manage. No webhooks to configure.

Install from Marketplace View source →
3
Git providers
8
AI hosting options
1
YAML block
$0
Infrastructure cost
Features

One task.
Everything you need.

Three actions — reviewPR, createPR, commentPR — covering the full pull request lifecycle from your pipeline.

🔍

AI code review

Claude reads your diff and posts a structured comment on the PR — covering bugs, security issues, performance, and style. Not a general suggestion: a real review.

�

Inline code suggestions

AI posts findings directly on changed lines with one-click fix suggestions. GitHub displays them as native ```suggestion blocks — developers can commit fixes instantly. Works on all platforms.

�📂

Per-file mode

Review each changed file individually, then synthesise findings into a consolidated PR assessment. Better signal on large, noisy diffs.

🌐

Multiple AI providers

Run on Anthropic direct, Azure AI Foundry, Azure OpenAI, AWS Bedrock, Google Vertex AI, Google AI Studio, GitHub Models, or your own LiteLLM proxy. Your infrastructure, your choice.

🔀

Create pull requests

Open a PR on any git host directly from your pipeline, with optional AI review on creation. Sets output variables for downstream steps.

📊

Output variables

Expose ReviewVerdict, ReviewTotalIssues, and ReviewSummary for downstream gates and notifications.

💬

Custom comments

Post build results, test coverage, or any markdown content on a PR. Comments are tagged so reviewers know what's human and what's automated.

Compatibility

Works with your stack,
not against it.

Connect to any major git host and run the model wherever your team already has credentials.

Git hosts

🐙 GitHub (cloud + Enterprise) 🦊 GitLab (cloud + self-hosted) 🪣 Bitbucket Cloud 🏢 Bitbucket Server / DC

AI providers

🟠 Anthropic (direct API) ☁️ Azure AI Foundry 🔷 Azure OpenAI Service 🟡 AWS Bedrock 🔵 Google Vertex AI 🌟 Google AI Studio 🐙 GitHub Models ⚡ LiteLLM proxy

Claude models

🚀 Haiku — fast, cost-effective ⚖️ Sonnet — recommended default 🔬 Opus — deepest analysis
How it works

Three steps.
No infrastructure.

Add the task to any existing pipeline and it just works.

1

Store your secrets once

Add your git host token and AI provider credentials to an Azure DevOps variable group. Mark them as secret. Done.

2

Drop the task in your pipeline

Add AiPrReviewer@1 to your YAML, set your provider, repository, and PR number. That's it.

3

Open a PR, get a review

The pipeline fetches the diff, sends it to Claude, and posts a structured review comment on your PR automatically.

In Action

Real reviews,
posted inline.

AI findings appear directly on changed lines in your pull request — with one-click fix suggestions developers can commit instantly.

AI Review HIGH security finding with suggested fix
🔒 HIGH · Security — Placeholder bundle ID detected in a production Terraform config. AI identifies the exact line and provides a runnable AWS CLI command to resolve it.
AI Review INFO best practice finding
📋 INFO · Best Practice — Backend config flags mismatch detected between code comments and the CI/CD pipeline. Suggests adding all required flags before merge.
AI Review MEDIUM best practice finding on diff line
💡 MEDIUM · Best Practice — HIPAA log retention below recommended value. Posted inline on the exact changed line with context-aware reasoning.
AI Review HIGH bug - runtime panic risk
🐛 HIGH · Bug — Runtime panic risk from missing bounds check in Terraform module. AI provides a complete precondition block as a suggested fix.
Pipeline Configurator

Build your pipeline YAML.
No guessing.

Pick your providers, action, and options — get a ready-to-paste snippet.

AI comments appear as native inline review comments with one-click ```suggestion fix blocks. Supported on GitHub, GitLab, and Bitbucket.

pipeline.yml 

        
      


    

  







  

    
Model Selection Guide

    
Choose the right AI model
for your scenario.

    
Different teams have different needs. Find your perfect match based on budget, quality, and scale.


    


      
      

        
🏆

        
Production Deployments

        
Security-critical code, high confidence needed

        
        

          
Claude Sonnet 4

          
⭐⭐⭐⭐⭐ Quality

        

        
        

          

            Cost/month:
            $140
          

          

            Context:
            200K tokens
          

          

            PRs/month:
            100
          

        

        
        

          
✓ Best for

          
    
            
  • Main branch deployments
    • 
            
  • Auth & payment code
    • 
            
  • Compliance reviews
    • 
          

        

      


      
      

        
💰

        
High-Volume Teams

        
200+ PRs/month, cost-conscious

        
        

          
GPT-4o

          
⭐⭐⭐⭐½ Quality

        

        
        

          

            Cost/month:
            $98
          

          

            Context:
            128K tokens
          

          

            Savings:
            30% cheaper
          

        

        
        

          
✓ Best for

          
    
            
  • Feature branches
    • 
            
  • Regular PRs
    • 
            
  • 90-95% Claude quality
    • 
          

        

      


      
      

        
🚀

        
Large PRs & Monorepos

        
1,000+ line PRs, architecture changes

        
        

          
Gemini 1.5 Pro

          
⭐⭐⭐⭐ Quality

        

        
        

          

            Cost/month:
            $49
          

          

            Context:
            2M tokens 🔥
          

          

            Savings:
            65% cheaper
          

        

        
        

          
✓ Best for

          
    
            
  • Massive refactors
    • 
            
  • 10,000+ line PRs
    • 
            
  • Cross-file reasoning
    • 
          

        

      


      
      

        
⚡

        
Draft/WIP PRs

        
Pre-review filtering, high volume

        
        

          
Gemini 1.5 Flash

          
⭐⭐⭐½ Quality

        

        
        

          

            Cost/month:
            $3
          

          

            Context:
            1M tokens
          

          

            Savings:
            98% cheaper
          

        

        
        

          
✓ Best for

          
    
            
  • Draft pull requests
    • 
            
  • Quick feedback
    • 
            
  • Speed > thoroughness
    • 
          

        

      


    


    
    

      
💵 Cost Comparison (100 PRs/month)

      
      

        

          
            

              Model
              Context
              Quality
              Small PRs
(5 files)		
              Medium PRs
(10 files)		
              Large PRs
(20 files)		
            

          
          
            

              Claude Sonnet 4 DEFAULT
              200K
              ⭐⭐⭐⭐⭐
              $105
              $140
              $280
            

            

              GPT-4o
              128K
              ⭐⭐⭐⭐½
              $72
              $98
              $196
            

            

              Gemini 1.5 Pro 2M CONTEXT
              2M
              ⭐⭐⭐⭐
              $36
              $49
              $98
            

            

              Gemini 1.5 Flash
              1M
              ⭐⭐⭐½
              $2
              $3
              $6
            

            

              DeepSeek V3
              64K
              ⭐⭐⭐⭐
              $7
              $10
              $20
            

          
        

      


      

        
💡 Pro Tip: Hybrid Strategy

        
Use Claude Sonnet for main branch (security-critical) and GPT-4o for feature branches. Save 50% while maintaining quality where it matters.

      

    


    
    

      
        📊 Full Model Comparison Guide
        
      
      

        Detailed analysis, real-world examples, migration guides & more
      

    


  







  

    
Examples

    
Copy. Change the repo name.
You're done.

    
Ready-to-use YAML for every provider combination.


    

      

        
GitHub

        
GitLab

        
Bitbucket

        
Azure AI Foundry

        
Azure OpenAI

        
AWS Bedrock

        
Google AI Studio

        
GitHub Models

      


      


trigger: none

pr:
  branches:
    include: [main]

variables:
- group: ai-reviewer-secrets   # GITHUB_PAT + ANTHROPIC_API_KEY

pool:
  vmImage: ubuntu-latest

steps:
- task: AiPrReviewer@1
  inputs:
    action:      reviewPR
    provider:    github
    accessToken: $(GITHUB_PAT)
    repository:  myorg/myrepo   # ← change this
    prNumber:    $(System.PullRequest.PullRequestNumber)
    enableAiReview: true
    aiApiKey:    $(ANTHROPIC_API_KEY)
    aiModel:     claude-sonnet-4-6

      


      


trigger: none

pr:
  branches:
    include: [main]

variables:
- group: ai-reviewer-secrets   # GITLAB_PAT + ANTHROPIC_API_KEY

pool:
  vmImage: ubuntu-latest

steps:
- task: AiPrReviewer@1
  inputs:
    action:      reviewPR
    provider:    gitlab
    accessToken: $(GITLAB_PAT)
    repository:  mygroup/myproject   # ← change this
        # serverUrl: https://gitlab.mycompany.com  # self-hosted only
    prNumber:    $(System.PullRequest.PullRequestNumber)
    enableAiReview: true
    aiApiKey:    $(ANTHROPIC_API_KEY)
    aiModel:     claude-sonnet-4-6

      


      


trigger: none

pr:
  branches:
    include: [main]

variables:
- group: ai-reviewer-secrets   # BITBUCKET_USERNAME + BITBUCKET_APP_PASSWORD + ANTHROPIC_API_KEY

pool:
  vmImage: ubuntu-latest

steps:
- task: AiPrReviewer@1
  inputs:
    action:      reviewPR
    provider:    bitbucket
    accessToken: $(BITBUCKET_USERNAME):$(BITBUCKET_APP_PASSWORD)
    repository:  myworkspace/myrepo   # ← change this
    prNumber:    $(System.PullRequest.PullRequestNumber)
    enableAiReview: true
    aiApiKey:    $(ANTHROPIC_API_KEY)
    aiModel:     claude-sonnet-4-6

      


      


steps:
- task: AiPrReviewer@1
  inputs:
    action:      reviewPR
    provider:    github
    accessToken: $(GITHUB_PAT)
    repository:  myorg/myrepo
    prNumber:    $(System.PullRequest.PullRequestNumber)
    enableAiReview: true
    aiProvider:  azure
    aiApiKey:    $(AZURE_AI_API_KEY)
    aiBaseUrl:   $(AZURE_AI_ENDPOINT)   # https://<resource>.services.ai.azure.com/models
    aiModel:     claude-sonnet-4-6       # your deployment name

      


      


steps:
- task: AiPrReviewer@1
  inputs:
    action:           reviewPR
    provider:         github
    accessToken:      $(GITHUB_PAT)
    repository:       myorg/myrepo
    prNumber:         $(System.PullRequest.PullRequestNumber)
    enableAiReview:   true
    aiProvider:       bedrock
    awsRegion:        us-east-1
    awsAccessKeyId:     $(AWS_ACCESS_KEY_ID)      # omit to use IAM role
    awsSecretAccessKey: $(AWS_SECRET_ACCESS_KEY)  # omit to use IAM role
    aiModel:          anthropic.claude-3-5-sonnet-20241022-v2:0

      


      


variables:
- group: ai-reviewer-secrets   # GITHUB_PAT + AZURE_OPENAI_API_KEY + AZURE_OPENAI_ENDPOINT

steps:
- task: AiPrReviewer@1
  inputs:
    action:      reviewPR
    provider:    github
    accessToken: $(GITHUB_PAT)
    repository:  myorg/myrepo
    prNumber:    $(System.PullRequest.PullRequestNumber)
    enableAiReview: true
    aiProvider:  azure-openai
    aiApiKey:    $(AZURE_OPENAI_API_KEY)
    aiBaseUrl:   $(AZURE_OPENAI_ENDPOINT)   # https://<resource>.openai.azure.com
    aiModel:     gpt-4o                    # your deployment name

      


      


variables:
- group: ai-reviewer-secrets   # GITHUB_PAT + GOOGLE_AI_API_KEY

steps:
- task: AiPrReviewer@1
  inputs:
    action:      reviewPR
    provider:    github
    accessToken: $(GITHUB_PAT)
    repository:  myorg/myrepo
    prNumber:    $(System.PullRequest.PullRequestNumber)
    enableAiReview: true
    aiProvider:  googleai
    aiApiKey:    $(GOOGLE_AI_API_KEY)       # from aistudio.google.com
    aiModel:     gemini-2.0-flash

      


      


variables:
- group: ai-reviewer-secrets   # GITHUB_PAT + GITHUB_MODELS_TOKEN

steps:
- task: AiPrReviewer@1
  inputs:
    action:      reviewPR
    provider:    github
    accessToken: $(GITHUB_PAT)
    repository:  myorg/myrepo
    prNumber:    $(System.PullRequest.PullRequestNumber)
    enableAiReview: true
    aiProvider:  githubmodels
    aiApiKey:    $(GITHUB_MODELS_TOKEN)     # GitHub PAT with models:read scope
    aiModel:     gpt-4o

      

    

  







  

    
Output variables

    
Use review results
in downstream steps.

    
Gate deployments, trigger Slack alerts, or log verdicts — all from the same pipeline.

    

      

        
PrUrl

        
Full URL to the pull request

      

      

        
PrNumber

        
Numeric PR / MR identifier

      

      

        
ReviewVerdict

        
lgtm · needs-work · critical

      

      

        
ReviewTotalIssues

        
Number of issues Claude found

      

      

        
ReviewSummary

        
One-line summary from the review

      

    

  







  

    

      
Start reviewing smarter.

      
Free to install. Runs on your existing Azure DevOps organisation. No new infrastructure.

      
        Install free from Marketplace