Hona.CodeReviewAi 0.0.4

Code Review AI 🤖

Tired of the copy-paste dance? You know the one: grab the PR title, description, hunt down the linked Jira ticket, copy its details, generate a diff, and then finally paste it all into your AI assistant for a code review? 😩

Yeah, that's tedious. This tool is the first step to automating that pain away.

What's the Point?

This little console app acts like your PR Context Butler. Give it an Azure DevOps Pull Request ID, and it does the legwork:

1. Fetches PR Details: Grabs the title and description from Azure DevOps.
2. Finds Linked Issues: Scans the PR description for Jira issue URLs (like https://your-co.atlassian.net/browse/TASK-123).
3. Gets Jira Context: Pulls the Jira issue's summary, description, comments, and even downloads any attachments right next to the output file!
4. Generates the Diff: Calculates the code changes between the source and target branches.
5. Bundles Everything: Creates a clean Markdown file (.md) containing all the gathered info, perfectly formatted for an LLM.

The Current Workflow (Semi-Manual)

Right now, it handles the gathering part. The review part still needs you (and your favorite AI):

1. Run the tool: code-review --pr <YourPrId>
2. It connects to AzDO & Jira, does its magic ✨.
3. It creates a folder like CodeReviews/PR-<YourPrId>/ containing:
   • PR-<YourPrId>-review.md: The main context file.
   • Any attachments downloaded from linked Jira issues.
4. Manual Step: Copy all files from the folder.
5. Paste that content into your AI chat (like t3.chat 😉, ChatGPT, Claude, etc.).
6. Send the prompt and wait for the review.

It saves a surprising amount of time and ensures you don't forget any crucial context!

Getting Started

1. Clone the repo.
2. Configure CodeReviewAi.Cli/appsettings.json:
   • AzureDevOps:
     • Organization: Your AzDO org name.
     • Project: Your AzDO project name.
     • RepositoryId: The ID or name of your repo.
     • PersonalAccessToken: Create an AzDO PAT with permissions:
       • Code > Read
       • Pull Request Threads > Read & write (needed for future auto-commenting)
   • Jira:
     • BaseUrl: Your Jira instance URL (e.g., https://your-co.atlassian.net).
     • User: Your Jira login email.
     • Token: Create a Jira API Token (from your account settings > security).
   • CodeReview:
     • OutputDirectory: Where to save the output folders (default: CodeReviews).
     • ReviewPrompt: The default text added to the end of the Markdown file.
     • IncludeUnchangedLinesInDiff: true for full-file diffs, false (default) for concise patch-style diffs.
3. Build: Build the CodeReviewAi.Cli project.
4. Install:
   • From source: Run install-dev.bat
   • From NuGet: dotnet tool install --global Hona.CodeReviewAi
5. Run:
   • code-review --pr <YourPrId>

What's Next? 🚀

This is just the MVP! The goal is full autonomy:

• Direct integration with LLMs (OpenAI-compatible APIs, maybe local models like Ollama).
• Automatically generating review comments.
• Posting those comments directly back to the Azure DevOps PR thread.
• Support for other providers (GitHub, GitLab, etc.).

Feel free to contribute or raise issues!

Template appsettings.json

{
  "AzureDevOps": {
    "Organization": "",
    "Project": "",
    "RepositoryId": "",
    "PersonalAccessToken": ""
  },
  "Jira": {
    "BaseUrl": "https://org.atlassian.net",
    "User": "",
    "Token": ""
  },
  "CodeReview": {
    "OutputDirectory": "CodeReviews",
    "ReviewPrompt": "Please review the above changes for correctness, clarity, and adherence to project standards. Highlight any issues, improvements, or questions below.",
    "IncludeUnchangedLinesInDiff": true
  }
}