An automated code quality agent that fetches your code, detects bugs using LLMs, and fixes them—but only when you say so.
The AI-Powered GitHub Debugger Agent is an n8n workflow designed to act as a "Human-in-the-Loop" DevOps bot. It autonomously audits your repository for bugs, inefficiencies, and redundancy.
Instead of blindly committing AI-generated code, it sends a detailed report to Discord. You review the changes, and if you approve (via a secure webhook link), the agent pushes the fixes directly to your repository.
📂 Auto-Fetch: Recursively scans your GitHub repository for code files (.js, .ts, .py, .cpp, etc.).
🧠 Intelligent Analysis: Uses GPT-4o to identify logic errors, O(n²) inefficiencies, and redundant variables.
🛡️ Human-in-the-Loop: Zero code is committed without your explicit click of an approval link.
🔔 Real-time Reporting: Delivers formatted Markdown bug reports directly to your Discord server.
⚡ Automated Fixes: Applies the AI-suggested refactor instantly upon approval.
The workflow operates in a linear 4-phase loop:
Ingestion: Fetches the file tree from GitHub and filters for source code.
Analysis: The LLM reads the code and outputs a JSON object containing a Documentation report and Fixed_Code.
Authorization: The workflow pauses and sends a webhook link to Discord.
Execution: Once the link is clicked, the workflow resumes and commits the changes via the GitHub API.
Prerequisites
n8n Instance: Self-hosted or Cloud (v1.0+).
GitHub Account: A Personal Access Token (Classic) with repo scope.
OpenAI API Key: Access to GPT-4o or GPT-3.5-Turbo.
Discord Server: Ability to create Webhooks.
Installation
Import the Workflow:
Copy the JSON workflow code (located in workflow.json or provided separately).
In n8n, go to Workflows > Import from JSON.
Configure Credentials:
GitHub API: Create a credential in n8n using your Personal Access Token.
OpenAI API: Create a credential using your API Key.
Setup Nodes:
Manual Trigger: Set your default Owner, Repo, and Branch.
Send to Discord: Paste your Discord Webhook URL.
Filter Code Files: (Optional) Add extensions like .rs or .go if needed.
Trigger the Agent:
Open the workflow in n8n and click "Execute Workflow" (or use the Manual Trigger form).
Wait for Analysis:
The agent will process files in batches.
Check Discord:
You will receive a notification: "🐞 Bug Report for src/app.js".
Approve or Ignore:
To Fix: Click the "Approve & Commit" link in the message.
To Skip: Simply ignore the message. The workflow will not modify the file.
We are constantly improving the agent. Here is what's coming in v2.0:
[ ] Pull Request Integration: Create a fix/ai-branch and open a PR instead of committing directly to main.
[ ] RAG (Context-Awareness): Use a vector store (Pinecone) so the AI understands cross-file dependencies.
[ ] Local LLM Support: Option to swap OpenAI for Ollama/Llama 3 for data privacy.
[ ] Compiler Loop: Run a build command (e.g., npm test) to verify the fix before asking for human approval.
Distributed under the MIT License. See LICENSE for more information.
Built with ❤️ using n8n and ☕