KnightLint: Your Code's Trusted Guardian

AI-powered pull request analysis that catches security flaws, code issues, and performance problems before they reach production.

Comment

Inspiration

Our inspiration for KnightLint came from seeing how time-consuming and inconsistent code reviews can be. Developers often rely on manual checks or scattered tools, leading to missed styling issues, exposed secrets (or security vulnerabilities), and poor code quality. We wanted to build a lightweight, automated linting companion that integrates directly into pull requests, making reviews cleaner, faster, and more secure. By turning tedious checks into a streamlined process, KnightLint empowers teams to focus on shipping better code.

What it does

1. Automated Code Review: Scans pull requests in real time to catch security vulnerabilities, code quality concerns, and performance issues before they are merged. 2. Pull Request Integration: Leaves clear, actionable comments directly on KnightLint website regarding the PRs so developers can fix issues without switching tools. 3. Team Collaboration: Standardizes review quality across teams, helping everyone focus on meaningful feedback instead of repetitive checks. 4. Lightweight & Developer-Friendly: Runs seamlessly alongside existing workflows, with zero extra setup, just log in with your Github account.

How we built it

Tech Stack: Backend: FastAPI (Python) for handling authentication, GitHub API communication, and automated linting tasks. GitHub Integration: Utilized the GitHub OAuth and Pull Request APIs to fetch repos and PRs. Frontend: React + TypeScript interface for displaying repositories, edit pull requests, and visualizing linting results. Linting Engine: Gemini AI checker that focuses on 1) security checks, 2) styling rules, and 3) secret scanning. It provides real time feedback of PR.

Key Implementation Details: Automated PR Workflow: Every pull request has an AI section that can be triggered to scan for security vulnerabilities, code quality concerns, and performance issues. Lightweight Setup: KnightLint is designed to integrate smoothly with existing repos. No extra configuration required from contributors. AI Re-checker & Committing Changes: Built-in option to commit changes after editing directly on KnightLint as well as option to recheck the PR after making relevant changes.

Challenges we ran into

Refining AI Response At first, the AI feedback on pull requests was inconsistent, too vague or overly strict. We spent time refining prompts, improving parsing, and tuning the logic to give developers clear, useful, and consistent feedback.

Deploying Website Connecting the frontend and backend during deployment was tricky. We ran into CORS issues, authentication callback errors, and environment variable setup problems. It took multiple rounds of debugging and testing to get a smooth production build.

Accomplishments that we're proud of

Seamless GitHub Integration: Built an automated linting system that uses GitHub APIs to fetch repository and pull request data, generate AI-powered feedback directly on KnightLint, and allow users to edit and commit changes back to the PR without leaving the platform.

Accurate & Useful AI Feedback: Refined the AI logic to detect styling issues, security risks, and secret exposures consistently, without overwhelming developers with noise.

Smooth Developer Experience: Created a lightweight setup that requires no extra steps from contributors. KnightLint is your one-stop-shop for successful PR completion.

What we learned

1. Prompt Refinement Matters: Fine-tuning how the AI responds to different code scenarios was key to making feedback clear, consistent, and actually helpful to developers.

2. Working with Gemini: Using Gemini to refine and structure the AI feedback taught us how powerful prompt iteration and structured context can be in producing developer-friendly responses. Especially since it was the first time everybody in the team worked with Gemini's API.

3. Deployment & Real-World Reliability: Getting the site to work seamlessly in production (not just locally) gave us a deeper understanding of CORS, environment configuration, and frontend-backend communication.

4. GitHub API Complexity: Integrating with GitHub OAuth and Pull Request APIs gave us hands-on experience with authentication flows, error handling, and secure request management.

What's next for KnightLint: Your Code's Trusted Guardian

Scalability & CI/CD Integration: Integrate with CI/CD pipelines to automate linting checks as part of deployments, ensuring clean and secure code makes it to production.

Team Collaboration Features: Introduce shared dashboards, review metrics, and PR activity timelines so teams can collaborate and improve their code review process together.

Customizable Rulesets: Give teams the power to define their own linting rules, severity levels, and security policies, making KnightLint flexible for any workflow.

Share this project:

Updates