OpenMaintainer — Autonomous AI Maintainer for Open Source

Built for NullShot Hacks: Season 0 (link)

Exploring the Agentic Web — powered by NullShot, Edenlayer, and Thirdweb

---

Important links

Frontend repo: https://github.com/GHkrishna/nullshot-open-maintainer-ui

Demo video: https://youtu.be/gH0pqsYY-2g

Brainstorm: https://nullshot.ai/brainstorm/9c32b729-05b9-4b5e-a822-49a562a91c96

Deployed contract: https://monad-testnet.socialscan.io/address/0xe35928e33d27325699d31513bedb818fc73cc795

---

Overview

OpenMaintainer is an AI-driven virtual maintainer for open-source repositories.
It reviews pull requests, might suggest improvements in the future, and automatically manages contributor rewards using transparent on-chain bounties.

By merging AI autonomy with Web3 verifiability, it demonstrates how the next generation of Agentic Systems can support open-source ecosystems.

---

Core Idea

Traditional open source relies on human maintainers — a scarce and overburdened resource.
OpenMaintainer introduces an AI Maintainer Agent that:

• Reviews PRs automatically (This might be pushed for further scope, or an agent like coderabbit can be used instead of having everything build from scratch)
• Suggests bounty rewards for contributions using AI evaluation
• Rewards contributors transparently using Thirdweb smart contracts

---

Key Components

Layer	Role	Tech used
AI Agent	Reviews PRs(future scope), analyzes contribution quality, and suggests bounty	NullShot TypeScript Agent Framework
Web3 Layer	Stores and distributes bounties	Thirdweb API + Solidity (Monad testnet)
Bridge	Links GitHub events ↔ AI logic ↔ blockchain calls	Node.js scripts
MCP	Enables discoverable and composable agent actions	Edenlayer / NullShot MCP API

---

Example Flow

1. Developer submits PR
2. AI Maintainer (via NullShot Agent) reviews PR and generates:
   • Summary
   • Review comments
   • Suggested bounty range (based on complexity, quality, and effort)
3. Maintainer approves → bounty auto-released via Thirdweb contract
4. Contributor receives reward tokens transparently on-chain

---

Token & Reward Mechanism

• Rewards are defined in bounty.md
• AI can update this dynamically based on contribution quality
• Smart contract ensures transparent distribution

---

Architecture

See architecture.md for a detailed breakdown.

High-level:

GitHub PR → NullShot AI Maintainer → AI Evaluation
     ↓
Suggested Bounty → Thirdweb Contract → On-chain Reward

Tech stack

Category	Technology
AI Agent	NullShot TypeScript Agent Framework
AI Model	LLM (Anthropic)
Smart Contract	Solidity + Remix + Thirdweb
Backend	Nestjs
Blockchain	Monad testnetc
Integration	GitHub REST API + Thirdweb API

Setup Instructions

Pre-requisite

• Github account and the access token with appropriate read write access to the repo
• Anthropic API keys
• Thirdweb API keys
• A deployed contrat on Monad testnet (with ERC20, tokens loaded in the generated address - see Contract section below)
• Github repo with appropriate tags, as specified in the Bounty.md file
• And a smile on ur face, coz this is a nobel cause that we would be using the solution for. (OSS is awesome 😉 )

Backend

git clone https://github.com/GHkrishna/nullshot-openmaintainer
cd nest-backend
cp .env.example .env
npm install
npm run dev

For the .env values refer the instructions specified in .env.example

Run AI maintainer:

cd ../agent

Add the following in .dev.vars and .env

AI_PROVIDER_API_KEY= your key
MODEL_ID= your anthropic model
AI_PROVIDER=anthropic

Run the agent

npm install
npm run dev

Frontend

git clone https://github.com/GHkrishna/v0-ai-maintainer-ui.git
cp .env.example .env
npm run dev

Contract

For now, you could simply deploy your own contract specified in the ./contracts folder

1. Get the address created by you on /on-chain/owner-account
2. Mint some token as the contract owner and send them to the created account

Hackathon Relevance

• NullShot Framework: used for building the AI agent
• Edenlayer MCP: ensures agent interoperability
• Thirdweb: handles smart contract deployment + token management

This project exemplifies the Agentic Economy — autonomous AI agents creating and distributing real economic value through verifiable, decentralized actions.

Future Scope

• Decentralized Agent Marketplace

  • Allow developers to create and publish their own AI agents for open source workflows.
  • Repositories can “plug and play” these agents (e.g., reviewer bot, issue triager, contributor engagement bot).

• Contributor Reputation & Insights

  • Introduce a scoring system to quantify impact beyond code — reviews, discussions, issue reports, etc.
  • These insights can help projects identify potential maintainers or reward genuine contributors.

• AI-Governed Bounty Management

  • Instead of manual or DAO voting, use AI-assisted reputation and contribution data to suggest and prioritize bounties automatically.
  • Later, a DAO layer can be added for decentralized validation and reward distribution.

• Multi-Agent Collaboration for Large Repositories

  • Enable multiple specialized agents (for docs, testing, CI/CD, etc.) to work together on large, multi-repo projects.
  • Agents can share context securely and coordinate to maintain consistency across repositories.