Open-Source Vulnerability Scanner for Cardano Smart Contracts

[GENERAL] Name and surname of main applicant

UnboundedMarket Team

[GENERAL] Are you delivering this project as an individual or as an entity (whether formally incorporated or not)

Entity (Not Incorporated)

[GENERAL] Please specify how many months you expect your project to last (from 2-12 months)

7

[GENERAL] Please indicate if your proposal has been auto-translated into English from another language

No

[GENERAL] Summarize your solution to the problem (200-character limit including spaces)

We develop an open-source AI model trained on Cardano smart contracts to identify potential bugs and vulnerabilities, reducing audit iterations and costs while improving security.

[GENERAL] Does your project have any dependencies on other organizations, technical or otherwise?

No

[GENERAL] If YES, please describe what the dependency is and why you believe it is essential for your project’s delivery. If NO, please write “No dependencies.”

No dependencies.

[GENERAL] Will your project’s output/s be fully open source?

Yes

[GENERAL] Please provide here more information on the open source status of your project outputs

Fully Open-Source under the MIT license.

[SOLUTION] Please describe your proposed solution

Problem:

Smart contract vulnerabilities pose a significant risk to the Cardano ecosystem, potentially leading to financial losses and reducing trust. Current auditing processes are time-consuming and costly and may miss critical bugs, hindering Cardano's adoption and growth.

Motivation: 

Previous studies have already demonstrated the potential for large language models to help identify bugs for other blockchains, such as Ethereum. Our approach builds on the findings of these studies and adapts them to Cardano.

Approach:

Our approach consists of three major steps:

Data Collection:

• Gather a wide range of Cardano smart contracts from various sources.
• Ensure diversity in the dataset by including contracts with different functionalitiess.
• Screen the dataset, and perform preprocessing and cleaning.

Model Fine-tuning:

• Test and select a suitable pre-trained LLM.
• Format and tokenize the dataset for fine-tuning
• Fine-tune the pre-trained LLM using the annotated dataset, optimizing it for the specific task of identifying bugs and vulnerabilities in Cardano smart contracts.
• Evaluate the fine-tuned model's performance.
• 
  

Development of user-friendly tools to run the model:

• Create user-friendly scripts that allow developers to easily interact with the fine-tuned model and perform audits on their smart contracts.
• Develop clear and concise documentation that guides users through the installation, setup, and usage.
• Provide examples and tutorials demonstrating how to run the model and interpret the model's outputs.

Expected Outcome: 

We expect that our project will engage various stakeholders within the Cardano community:

• Developers: The AI auditing tool will assist developers early in the development process, enhancing security and reducing time-to-market by identifying potential vulnerabilities and strengthening the security of their smart contracts.
• Auditors: The tool will complement traditional auditing processes, providing an additional layer of analysis.
• Users: The AI auditing tool will enhance the security of smart contracts, protecting users' assets and increasing confidence in the Cardano platform.
• Researchers: The project's open-source nature will encourage collaboration and knowledge sharing among researchers interested.

[IMPACT] Please define the positive impact your project will have on the wider Cardano community

Our open-source AI model for auditing Cardano smart contracts will have a significant positive impact on the Cardano community in several ways:

1. Enhanced security: By identifying potential bugs and vulnerabilities, the AI model will help improve the overall security of smart contracts deployed on the Cardano blockchain, protecting users' assets and reducing the risk of exploits.
2. Cost reduction: The AI model directly benefits developers by reducing the number of re-audit iterations required. The model can identify issues before a separate entity audits the smart contract, lowering the overall cost of smart contract development and auditing and making it more accessible for developers to create and deploy secure smart contracts on Cardano.
3. Increased adoption: With improved security and reduced costs, more developers are encouraged to build on Cardano, leading to increased adoption and growth of the ecosystem.
4. Trust and confidence: The AI model can significantly enhance trust and confidence in the Cardano blockchain. Providing an additional layer of security can attract more users and investors to the platform.
5. Community collaboration: As an open-source project, the AI model will foster collaboration within the Cardano community, allowing developers to contribute, improve, and adapt the model to suit their needs.

[CAPABILITY & FEASIBILITY] What is your capability to deliver your project with high levels of trust and accountability? How do you intend to validate if your approach is feasible?

Our team consists of four AI experts and blockchain enthusiasts with proven track records in academia and crypto development and up to 8 years of experience in relevant fields. Our work-philosophy is best described by:

Open: Achievements and work artifacts are open-sourced and published on a regular basis.

 

Transparent: Using clear documentation and academic rigor, we build trust and accountability. Whenever appropriate, we are happy to discuss our work with the academic community at top-tier blockchain and AI conferences around the globe.

Collaborative: Through collaboration with established industry-peers from our earlier blockchain ventures, we gain regular feedback to continuously improve and adjust where required.

[PROJECT MILESTONES] What are the key milestones you need to achieve in order to complete your project successfully?

Research and Planning (2 months)

Milestone outputs:

• Conduct a comprehensive literature & market review on previous studies and existing automated tools for finding bugs in smart contracts on other blockchains.
• Identify specific requirements and challenges for Smart Contracts on Cardano.
• Develop a detailed project plan, including timelines, resource allocation, and deliverables.

Acceptance criteria:

The final research report covers the important aspects of the Milestone’s outputs, including:

• Literature & Market Review
• Proposed solution
• Project Plan

Evidence of milestone completion:

• A detailed research report summarizing all findings from the literature & market review, outlining the proposed solution, and presenting the project plan.

Data Collection and Preprocessing (2 months)

Milestone outputs:

• Collect a diverse Cardano smart contracts' dataset, including secure and vulnerable examples.
• Clean and preprocess the dataset, ensuring consistency and quality.
• Establish a data pipeline for continuous dataset updates and maintenance.
• Open-source all data and code used to collect and preprocess data.
• Release a comprehensive documentation of the whole process and the code.

Acceptance criteria:

The open-source GitHub Repository contains:

• The curated dataset
• The documented code used for all preprocessing steps

Evidence of milestone completion:

• An open-source GitHub repository containing the code, data, and documentation

Model Development and Fine-tuning (2 months)

Milestone outputs:

• Select an appropriate pre-trained language model as the foundation for the AI auditing tool.
• Fine-tune the model on the collected Cardano smart contract dataset.
• Conduct testing and benchmarking of the model's performance and accuracy.
• Develop additional scripts and examples, so the community can easily use the model.
• Write detailed documentation for the code, including instructions on how to set up environments and run the code.

Acceptance criteria:

The open-source GitHub repository is updated and includes:

• Links to the fine-tuned model.
• Documentation of the model selection, fine-tuning, and testing.
• Documented code and instructions on how to set up the environment and run the code.

Evidence of milestone completion:

• A link to the repository with the updated code fulfilling the acceptance criteria

Iterative Feedback & Improvements (1 month)

Milestone outputs:

• Engage with the community to gather feedback and suggestions.
• Iteratively improve the model, scripts, and examples based on feedback.
• Extensively document changes and updates.

Acceptance criteria:

• The final report summarizes all steps, including feedback received and documentation of any changes or fixes.

Evidence of milestone completion:

• The final closeout-report and video

[RESOURCES] Who is in the project team and what are their roles?

Timo is a visionary entrepreneur with a deep expertise in robotics, artificial intelligence, and cloud engineering. He holds a Master’s degree in Robotics with a focus on AI applications and co-founded an AI-driven agricultural startup. Timo has published his work at a major conference, demonstrating his ability to merge academic research with real-world solutions. Timo is driven by his passion to unlock the pioneering potential in bringing AI solutions to the blockchain.

Samy is a professional with five years of blockchain development experience and nearly eight years in AI and machine learning. As a developer and researcher, Samy has worked on several DeFi protocols and published peer-reviewed articles in top-tier AI journals and conferences.

Chris is a seasoned software developer and entrepreneur, boasting over four years of specialized experience in blockchain and AI technologies. His in-depth research in AI and blockchain testifies his comprehensive knowledge of the DeFi ecosystem, establishing him as a knowledgeable figure within the community.

Phil is a former AI & Data scientist with a proven academic track record and several years of industrial experience. His professional career revolved around the robotics industry, where he successfully demonstrated the use of AI to tackle challenging problems. As a crypto-enthusiast, he now seeks to contribute to the next wave of blockchain innovation.

[BUDGET & COSTS] Please provide a cost breakdown of the proposed work and resources

Cost Breakdown:

Research and Planning (18,000 ADA):

• Literature review and analysis of existing auditing tools: 3,000 ADA
• Identification of Cardano-specific requirements and challenges: 2,000 ADA
• Project planning, resource allocation, and deliverable definition: 5,000 ADA
• Research report and documentation: 8,000 ADA

Data Collection and Preprocessing (21,000 ADA):

• Cardano smart contract dataset collection: 9,000 ADA
• Dataset cleaning and preprocessing: 7,000 ADA
• Data pipeline development for continuous updates: 5,000 ADA

Model Development and Fine-tuning (32,000 ADA):

• Pre-trained language model selection and model fine-tuning on Cardano smart contract dataset: 16,000 ADA
• Extensive testing and benchmarking of the model's performance: 8,000 ADA
• Develop Scripts, examples and write extensive documentation: 8,000 ADA

Improvements, Feedback Integration and Community Engagement (16,000 ADA):

• Obtain community feedback and suggestions. 2,000 ADA
• Iteratively improve the model, scripts, and examples based on feedback. 10,000 ADA 
• Extensively document changes and updates. 4,000 ADA

Reserve for unforeseen costs: 7,000 ADA

Total Cost: 93,000 ADA

[VALUE FOR MONEY] How does the cost of the project represent value for money for the Cardano ecosystem?

The cost of developing an open-source AI-powered auditing tool for Cardano smart contracts represents excellent value for money for the Cardano ecosystem due to its far-reaching benefits and long-term impact:

1. Cost savings for developers: By reducing the number of re-audit iterations and streamlining the auditing process, the AI tool has the potential to significantly lower the costs associated with smart contract development and deployment
2. Enhanced security and trust: The AI auditing tool will provide an additional layer of security, identifying potential vulnerabilities and bugs that traditional auditing methods may miss. This improved security will protect users' assets, increase trust in the Cardano platform, and attract more users and investors to the ecosystem. The value of preventing even a single major exploit can far outweigh the cost of developing the tool.
3. Community-driven development: The project's open-source nature ensures that the AI auditing tool will be a community-owned asset. This approach maximizes value for money by leveraging the collective knowledge and expertise of the Cardano community.