gitDontIgnore

An AI-powered onchain project manager that handles the software development flow—from idea to delivery—without traditional human coordination. As a client, you start by verifying your identity and explaining your project to an AI agent. The agent understands your needs and creates a GitHub repository, setting the stage for development. It then opens the project for bidding, allowing developers to apply. Each developer's GitHub is analyzed by an AI agent, which scores their experience and their proof is verifies onchain and onboarded onto the project. Once selected, the developer gets added as a collaborator and gets a dedicated branch to work on. After completing their module, the developer submits his module for review, where the AI Agent tests the feature in an ai automated browser. If everything works as expected, it releases funds to the developer, opens a pull request, and merges the work to the main branch. If not, it raises a GitHub issue and provides feedback. Throughout the process, AI agents handle project management, quality control, and fund distribution, while all transactions and interactions are recorded onchain in the celo network. The goal is to make project execution smooth, and efficient powered by onchain ai agents and zk proofs.

The project’s core is the Self onchain proof system, which uses zero-knowledge proofs for secure identity verification, allowing clients and developers to onboard privately without revealing sensitive data. We also integrated GitHub verification onchain to assess and score developers based on their GitHub profiles, ensuring only qualified individuals are selected for the project.

A key component of the project was the development of two custom plugins for Vercel and GitHub within the Goat SDK. These plugins allowed us to automate interactions with GitHub repositories and manage deployments on Vercel, bridging the AI agents with project workflows like repository creation, code pushes, and pull requests.

We deployed the project’s smart contracts on the Celonetwork, ensuring that all transactions and fund releases were secure, scalable, and eco-friendly. For module testing, we used the Anchor AI browser to automatically test the code in a controlled AI environment, ensuring everything works before funds are released.

A notable aspect was the creation of the custom plugins, which allowed seamless integration between different services and fully automated tasks like testing, fund release, and project management.