🕵️ ChangeReviewLogger: Ask an AI agent to review a code change

ChangeReviewLogger is an AI-powered GitHub App that listens to your pull request review comments and replies to an "AnalyseThis" request with a report detailing out the reasons and impacts of that particular code change.

🚀 How It Works

Every time a new review comment is added to a PR:

• The bot looks for the term "AnalyseThis" (CASE SENSITIVE)
• If found, it generates a Summary of the changes (Pros, Cons, Reasons etc) and scores the change out of 100.
• 100 means highly confident about making this change.
• 0 means dont approve this change, please.

📍 Where to Find the report:

• As a reply to the review comment.

🧩 Project Overview

This repository runs two integrated services via Docker Compose:

1. MonitoringBot – a GitHub ProBot-based automation bot (/monitoring_bot).
2. Assistant – a CrewAI-based LLM agent (/assistant).

Both services rely on .env configuration files for secrets and environment variables. These files are excluded from version control for security reasons.

---

🛠️ Prerequisites

Before you begin, ensure you have:

• Docker and Docker Compose installed and running.
• A valid Gemini API key for the Assistant. [the API key is free for use]
• Create two empty .env files, one in assistant and other in monitoring_bot directories.

---

🧠 Configure the Assistant (LLM Service)

Create a .env file in the assistant/ directory using the following template:

# LLM model identifier (e.g., Gemini Flash)
MODEL=gemini/gemini-2.0-flash-001


# API key for your LLM provider
GEMINI_API_KEY=YOUR_GEMINI_API_KEY

---

🚀 Launch with Docker Compose

The root of the project contains a docker-compose.yml that references both services and automatically loads each .env file from the respective folders.

To build and start the services:

docker-compose up --build

---

🚀 Launch with Docker Compose

The root of the project contains a docker-compose.yml that references both services and automatically loads each .env file from the respective folders.

To build and start the services:

docker-compose up --build

---

🕹️ Configure the GitHub Monitoring Bot

Once the services are up, the console will display a local URL like:

http://localhost:3000

👉 Follow These Steps in Your Browser:

• Open the URL above.

• Register a new GitHub App through the interface.

• During setup, select the repositories you want the bot to monitor.

• Save the configuration.

• ℹ️ This process will automatically populate environment variables inside the .env file in the monitoring_bot/ directory.

• 🔄 Restart the Server After the .env file is updated, restart the server for changes to take effect:

docker-compose down
docker-compose up

---

🔐 Required GitHub App Permissions (Verification step, in case the end to end flow does not work)

Ensure the following permissions and event subscriptions are defined in your app.yml (used for GitHub App manifest creation):

default_events:
  - commit_comment
  - issue_comment
  - issues
  - pull_request
  - pull_request_review
  - pull_request_review_comment
  - push
  - repository

default_permissions:
  issues: write
  metadata: read
  pull_requests: write
  statuses: write

Example results: