This is a submission for the GitHub Copilot Challenge: New Beginnings
What I Built
Every developer has felt that moment of hesitation before sharing their code. That flutter of uncertainty, wondering if their work will measure up. As I sipped my coffee and opened VS Code, I imagined a softer, kinder way to conduct reviews—an environment more like a patient conversation with a friend rather than a cold scrutiny.
That vision is Pear Review: a code reviewer that meets you where you are, offering gentle wisdom and real-time insights. Built in a whirlwind 24-hour session, Pear Review harnesses GitHub Copilot and VS Code's extension APIs to instantly comment on your code changes, turning each development step into an opportunity for growth and clarity.
The Challenge
The biggest hurdle was ensuring that the feedback didn’t feel clinical or robotic. Traditional reviews often:
- Catch issues too late
- Cause anxiety over potential judgment
- Skip essential teachable moments
- Feel detached from the immediate editing process
On the technical side, integrating Git-based change detection and diagnostic reporting in VS Code demanded careful orchestration. I had to ensure that each file’s modifications were captured, processed, and annotated in a friendly, empathetic manner.
The Solution
Pear Review brings empathy to code reviews by blending precise diagnostics with a supportive tone. This required a pipeline that:
- Scans for changes in Git, to identify new or modified files.
- Sends each changed snippet to the Language Model API
- Populates VS Code’s Diagnostics with review comments.
- Displays them instantly, leaving you free to accept or ignore suggestions without ever leaving your editor.
The result is a partner in your editor, ever ready to offer suggestions, encouragement, and empathy.
Demo
Seeing Pear Review in action is often the “aha!” moment. As you write code, you’ll notice the 🍐 icon happily waiting in the status bar. If you prefer manual reviews, simply click the pear to receive immediate, insightful feedback. But if you want Pear Review to always have your back, toggle “Review on Save.” This enables automatic reviews every time you save, with quick checks and inline diagnostic comments that are easy to absorb.
Installation
You can download this extension directly from the Visual Studio Marketplace:
VS Code Marketplace – Pear Review
Repo
For a closer look, visit the official repository:
GitHub: Pear Review
Copilot Experience
Building Pear Review in 24 hours meant leveraging every available tool—chief among them, GitHub Copilot. Its ability to generate scaffolding, comment templates, and draft and edit code accelerated the process dramatically. From setting up a minimal ‘CHANGELOG.md’ for versioning to building out the core review logic, Copilot’s assistance ensured efficiency without compromising quality.
Initial Steps with Copilot
Before any code was written, Copilot guided the basic file structure, bundling essential TypeScript definitions and helping me configure the extension’s main entry points.
Workflow
As soon as I got the extension skeleton running, Copilot was there providing real-time suggestions—often anticipating the next step in hooking up the Git tracking or shaping the review flow.
Model Switching for Different Tasks
A single model wasn’t always best for every context. Sometimes, GPT 4o handled short suggestions, while deeper logic drafts benefited from Claude 3.5 Sonnet. Swift switching made for agile development and fewer context resets.
Documentation Creation
The README.md practically wrote itself with Copilot’s help, ensuring every new feature got a proper mention. Meanwhile, CONTRIBUTING.md borrowed Copilot’s knack for clarity, encouraging community standards that keep empathy at the forefront.
VS Code Integration
Tying it all together, Copilot even suggested how to utilize the vscode.languages.createDiagnosticCollection in the review logic, letting comments display right where they matter most: in the code editor.
Marketplace Publication
Finally, Copilot offered a handy checklist for publishing to the VS Code Marketplace. This minimized last-minute snags, allowing me to polish the extension’s branding and get Pear Review ready for curious coders.
Key Copilot Features Used
- Edits for refining code
- Real-time Chat to clarify tricky APIs
- Model Switcher for context-appropriate analyses
- Inline Suggestions for faster iteration
Conclusion
Through late-night debugging sessions and sunlit morning code polishes, Pear Review became more than a project—it became a testament to how empathy can coexist with robust functionality. By treating each line of code as a learning moment and each comment as a chance to connect, we’ve imbued Pear Review with something that’s often missing in automated tools: heart.
In 24 hours, I witnessed how quickly an idea can bloom when nurtured by AI insights and personal passion. Pear Review stands as a friendly nudge toward a future where code reviews enable confidence, foster mutual respect, and feel a bit like sharing recipes with a caring neighbor—not a dreaded academic exam.
Thank you for reading, and may your coding journeys be filled with supportive reviews and sweet, pear-fect insights! 🍐
Top comments (1)
Wow!! This is a very thoughtful implementation of AI. I loved how you made use of AI, not only making it a clever project but also a tool that transforms potentially stressful reviews into gentle ones. Great job!