An advanced AI-driven SAST engine that utilizes Tree-sitter for AST parsing, Microsoft CodeBERT for semantic featurization, and PyTorch Geometric GCNs for static code vulnerability detection, paired with a local DeepSeek LLM triage assistant loop.
Python
1
14 commits
updated Jul 16, 2026
An Advanced Graph Neural Network (GNN) based Static Application Security Testing (SAST) Tool
GNNspector is a next-generation AI-driven static analysis tool engineered to detect vulnerabilities (such as Buffer Overflows, Command Injections, etc.) in source code. Instead of relying on traditional regular expressions (Regex), it parses source code into structured Abstract Syntax Trees (AST) using Tree-sitter, generates deep semantic node features via Microsoft CodeBERT, and utilizes a custom Graph Neural Network (GCN) architecture to classify files as Safe or Vulnerable.
Post-detection, it routes the telemetry to a hybrid LLM core (Cloud / Local Ollama) to generate a full remediation report and initiates an interactive security assistant patch loop directly inside your terminal.
To ensure the scanner operates correctly, your environment must satisfy the following system and software requirements:
The engine utilizes Ollama for hosting local LLMs to guarantee data privacy during code audits.
Run this single command to download and install Ollama automatically:
curl -fsSL https://ollama.com/install.sh | sh
To run the model, use:
ollama run deepseek-coder:1.3b
OllamaSetup.exe file and click Install.Follow these steps to set up the environment and install GNNspector:
# Update system and install dependencies
sudo apt-get update
sudo apt-get install build-essential python3-dev
# Clone the repository and navigate inside
git clone https://github.com/bkshukla91/GNNspector.git
cd GNNspector
# Install Python requirements
pip install -r requirements.txt
pip install -e .
# [OPTIONAL] Set up cloud API key if local Ollama/DeepSeek is not working
OPENROUTER_API_KEY="your_actual_api_key_here"
# Run GNNspector
gnnspector path/to/your/target_code.c
This architecture is distributed under the MIT License - see the LICENSE file for open-source compliance details.
An advanced AI-driven SAST engine that utilizes Tree-sitter for AST parsing, Microsoft CodeBERT for semantic featurization, and PyTorch Geometric GCNs for static code vulnerability detection, paired with a local DeepSeek LLM triage assistant loop.
Python
1
14 commits
updated Jul 16, 2026
An Advanced Graph Neural Network (GNN) based Static Application Security Testing (SAST) Tool
GNNspector is a next-generation AI-driven static analysis tool engineered to detect vulnerabilities (such as Buffer Overflows, Command Injections, etc.) in source code. Instead of relying on traditional regular expressions (Regex), it parses source code into structured Abstract Syntax Trees (AST) using Tree-sitter, generates deep semantic node features via Microsoft CodeBERT, and utilizes a custom Graph Neural Network (GCN) architecture to classify files as Safe or Vulnerable.
Post-detection, it routes the telemetry to a hybrid LLM core (Cloud / Local Ollama) to generate a full remediation report and initiates an interactive security assistant patch loop directly inside your terminal.
To ensure the scanner operates correctly, your environment must satisfy the following system and software requirements:
The engine utilizes Ollama for hosting local LLMs to guarantee data privacy during code audits.
Run this single command to download and install Ollama automatically:
curl -fsSL https://ollama.com/install.sh | sh
To run the model, use:
ollama run deepseek-coder:1.3b
OllamaSetup.exe file and click Install.Follow these steps to set up the environment and install GNNspector:
# Update system and install dependencies
sudo apt-get update
sudo apt-get install build-essential python3-dev
# Clone the repository and navigate inside
git clone https://github.com/bkshukla91/GNNspector.git
cd GNNspector
# Install Python requirements
pip install -r requirements.txt
pip install -e .
# [OPTIONAL] Set up cloud API key if local Ollama/DeepSeek is not working
OPENROUTER_API_KEY="your_actual_api_key_here"
# Run GNNspector
gnnspector path/to/your/target_code.c
This architecture is distributed under the MIT License - see the LICENSE file for open-source compliance details.