@parvez3019/ai-code-review-gitlab-plugin

Based on - https://github.com/hataiit9x/gemini-review-code

Summary

@parvez3019/ai-code-review-gitlab-plugin It is a small tool used for inline code review in GitLab Merge Requests. It supports calling the GitLab API for adding inline review comments and uses either Gemini AI API or AWS Bedrock to obtain AI code review feedbacks. Please note that when using it, ensure compliance with company regulations. 😉

Features

• 🛠️ Support configuration GitLab API address.
• ⚙️ Support configuration AWS Bedrock (Claude or any bedrock models) for code review.
• 📦 Support configuration GitLab Project ID & Merge Request IID.
• 🚀 Support running in Gitlab CI/CD Pipeline.
• 🚦 Automatically wait and try again when the rate limit is exceeded.
• 💬 The review results are appended to the location of the corresponding code block in the form of comments.
• 🔒 Secure handling of AWS credentials and GitLab tokens.
• 🏷️ Easy integration with GitLab CI/CD pipelines.
• 📝 Provide Summary Sections for the changes where no feedback is required.
• 📝 Customizable prompts via local files or S3

Screenshots for comments on Gitlab Merge Requests

Review Comment:

Summary Comment For No Feedback Section -

Install

1npm i @parvez3019/ai-code-review-gitlab-plugin

Environment Variables

AWS Bedrock Configuration

1# Required for AWS Bedrock
2export AWS_REGION="us-east-1"
3export AWS_SECRET_ACCESS_KEY="your-aws-secret-key"
4export AWS_ACCESS_KEY_ID="your-aws-access-key"
5# Available Claude models in Bedrock:
6# - anthropic.claude-3-5-sonnet-20241022-v2:0
7export AWS_BEDROCK_MODEL="anthropic.claude-3-5-sonnet-20241022-v2:0"

Gemini Configuration

1# Required for Gemini
2export GEMINI_API_KEY="your-gemini-api-key"
3export GEMINI_API_URL="https://generativelanguage.googleapis.com"
4export GEMINI_MODEL="gemini-1.5-flash"

General Configuration

1# Optional: Set default AI provider
2export AI_PROVIDER="bedrock"  # or "gemini"

Usage

Command Line Arguments

1ai-code-review-gitlab-plugin \
2  --gitlab-api-url "https://gitlab.com/api/v4" \
3  --gitlab-access-token "your-gitlab-token" \
4  --project-id 123 \
5  --merge-request-id 456 \
6  --ai-provider bedrock \  # or gemini
7  --api-key "your-api-key" \
8  --api-secret "your-api-secret" \  # required for bedrock
9  --region "us-east-1" \  # for bedrock
10  --custom-model "model-id" \  # optional
11  --system-prompt-path "/path/to/system-prompt.txt" \  # optional, local file or s3://bucket/key
12  --code-review-prompt-path "/path/to/code-review-prompt.txt" \  # optional, local file or s3://bucket/key
13  --s3-region "us-east-1" \  # optional, for S3 prompt paths
14  --s3-access-key "your-s3-access-key" \  # optional, for S3 prompt paths
15  --s3-secret-key "your-s3-secret-key"  # optional, for S3 prompt paths

GitLab CI/CD Integration

Add the following to your .gitlab-ci.yml file to enable automated code reviews:

1code-review:
2  stage: code-review
3  image: node:20
4  script:
5    - npm i -g @parvez3019/ai-code-review-gitlab-plugin
6    - |
7      if [ "$AI_PROVIDER" = "bedrock" ]; then
8        ai-code-review-gitlab-plugin \
9          -t "$CODE_REVIEW_GITLAB_TOKEN" \
10          -p "$CI_MERGE_REQUEST_PROJECT_ID" \
11          -m "$CI_MERGE_REQUEST_IID" \
12          -a $AI_PROVIDER \
13          -k "$AWS_ACCESS_KEY_ID" \
14          -s "$AWS_SECRET_ACCESS_KEY" \
15          -r "$AWS_REGION" \
16          -c "$AWS_BEDROCK_MODEL" \
17          -sp "$SYSTEM_PROMPT_PATH" \
18          -crp "$CODE_REVIEW_PROMPT_PATH" \
19          --s3-region "$S3_REGION" \
20          --s3-access-key "$S3_ACCESS_KEY" \
21          --s3-secret-key "$S3_SECRET_KEY"
22      else
23        ai-code-review-gitlab-plugin \
24          -t "$CODE_REVIEW_GITLAB_TOKEN" \
25          -p "$CI_MERGE_REQUEST_PROJECT_ID" \
26          -m "$CI_MERGE_REQUEST_IID" \
27          -a $AI_PROVIDER \
28          -k "$GEMINI_API_KEY" \
29          -c "$GEMINI_MODEL" \
30          -sp "$SYSTEM_PROMPT_PATH" \
31          -crp "$CODE_REVIEW_PROMPT_PATH" \
32          --s3-region "$S3_REGION" \
33          --s3-access-key "$S3_ACCESS_KEY" \
34          --s3-secret-key "$S3_SECRET_KEY"
35      fi
36  only:
37    - merge_requests
38  when: manual
39  tags:
40    - build
41    - stg

Alternative Trigger Methods

1. Using GitLab Webhooks

You can set up automatic code reviews when you are added as a reviewer using GitLab Webhooks:

1. Go to your GitLab project settings

2. Navigate to Webhooks

3. Add a new webhook with the following settings:

   • URL: Your webhook endpoint (e.g., AWS Lambda function URL)
   • Trigger: Select "Merge request events"
   • Secret token: Generate a secure token

4. Create a webhook handler (example using AWS Lambda):

1exports.handler = async (event) => {
2    const payload = JSON.parse(event.body);
3    
4    // Check if this is a merge request event
5    if (payload.object_kind === 'merge_request') {
6        // Check if the action is adding a reviewer
7        if (payload.action === 'reviewer') {
8            // Check if you are the added reviewer
9            const yourGitlabId = 'YOUR_GITLAB_USER_ID';
10            const addedReviewers = payload.changes.reviewers?.current || [];
11            
12            if (addedReviewers.some(reviewer => reviewer.id === yourGitlabId)) {
13                // Trigger the code review
14                const response = await fetch('https://gitlab.com/api/v4/projects/' + 
15                    payload.project.id + '/merge_requests/' + 
16                    payload.object_attributes.iid + '/pipeline', {
17                    method: 'POST',
18                    headers: {
19                        'PRIVATE-TOKEN': process.env.GITLAB_TOKEN
20                    }
21                });
22            }
23        }
24    }
25    
26    return {
27        statusCode: 200,
28        body: JSON.stringify('Webhook processed')
29    };
30};

This setup will:

1. Listen for merge request events
2. Check if you are added as a reviewer
3. Automatically trigger the code review pipeline when you are added

Make sure to:

• Replace YOUR_GITLAB_USER_ID with your actual GitLab user ID
• Set up the necessary environment variables in your webhook handler
• Configure proper security measures for your webhook endpoint

2. Using GitLab API Directly

You can also trigger the code review directly using the GitLab API:

1curl --request POST \
2  --header "PRIVATE-TOKEN: $GITLAB_TOKEN" \
3  "https://gitlab.com/api/v4/projects/$PROJECT_ID/merge_requests/$MR_IID/pipeline"

This can be integrated into your existing workflows or automation scripts.

Options

Option	Description	Default	Required
-g, --gitlab-api-url	GitLab API URL	https://gitlab.com/api/v4	No
-t, --gitlab-access-token	GitLab Access Token	-	Yes
-p, --project-id	GitLab Project ID	-	Yes
-m, --merge-request-id	GitLab Merge Request ID	-	Yes
-a, --ai-provider	AI Provider (gemini or bedrock)	gemini	No
-k, --api-key	API Key (Gemini API Key or AWS Access Key ID)	-	Yes
-s, --api-secret	API Secret (AWS Secret Access Key for Bedrock)	-	Yes (for bedrock)
-r, --region	AWS Region for Bedrock	us-east-1	No
-c, --custom-model	Custom Model ID	-	No
-sp, --system-prompt-path	System Prompt Path	-	No
-crp, --code-review-prompt-path	Code Review Prompt Path	-	No
-sr, --s3-region	S3 Region	-	No
-sk, --s3-access-key	S3 Access Key	-	No
-ssk, --s3-secret-key	S3 Secret Key	-	No

Troubleshooting

AWS Bedrock Issues

1. Model Not Available Error

   • Make sure you're using a supported model ID
   • Check if you have access to the model in your AWS region
   • Verify that your AWS account has the necessary permissions

2. Authentication Errors

   • Verify your AWS credentials are correct
   • Check if your AWS credentials have the necessary permissions for Bedrock
   • Ensure your AWS region is correct

3. Rate Limiting

   • The tool will automatically retry when rate limited
   • Consider using a different model if you're consistently hitting rate limits

Gemini Issues

1. API Key Issues

   • Verify your Gemini API key is correct
   • Check if your API key has the necessary permissions
   • Ensure you're using the correct API URL

2. Model Availability

   • Make sure you're using a supported model ID
   • Check if your API key has access to the specified model

Requirements

• Node.js >= 20.0.0
• GitLab account with API access
• Google Cloud account (for Gemini) or AWS account (for Bedrock)

Contributing

Welcome to contribute code, ask questions and suggestions! 👏

License

This project is based on the MIT license. See the LICENSE file for details. 📜