Cha-Ching

Cha-ching is a decentralized bounty platform for GitHub issues. It solves one of open source's core problems: contributors keep the internet running, but most of them work for free.

With Cha-ching, maintainers can post bounties (in crypto) directly on GitHub issues. Developers (bounty hunters) find the issues, fix them, and submit pull requests. Once a PR is merged, Cha-ching automatically handles the payout through a smart contract — no signups, no wallet connection, no friction.

This system brings sustainability to OSS by embedding compensation directly into contribution workflows, which not only helps attract and retain valuable contributors but also elevates the overall quality and reliability of open-source projects directly into the contribution workflow. The protocol uses smart contracts for escrow and payment, GitHub webhooks for issue+PR tracking, and AI agents to suggest relevant bounties to contributors by analyzing their coding language preferences and natural language prompts. It’s fast, fun, and finally fair to developers.

Tech Stack:

• Smart Contracts: Solidity-based contracts for escrow and payout, deployed on Optimism for fast, low-cost execution.

• Vlayer: for generating Web2 proofs from GitHub data — for example, data from the Pull Request, if it was merged, etc.

• Backend: Built with NestJS, it listens to GitHub webhooks (pull_request, issues, issue_comment, etc.) and triggers logic for PR validation and payments (on-chain).

• Frontend: A super intuitive React + Tailwind UI lets maintainers post bounties and track status. Bounty hunters don’t need to sign-up to the platform. They just see the available issues in Cha-Ching and get the recommendation of which ones they are likely skilled to do it.

• A.I. + Automation:

  • We use OpenAI plus custom MCP (Model Context Protocol) servers for MongoDB.
  • The GitHub Agent evaluate PR events, handle edge cases, and verify connections between PRs and bountied issues.
  • Moreover, our agent also provides a match between bounty hunters to potentially suitable bounties they fit based on programming language interests and a natural language input field.

• Database: MongoDB, accessed via an MCP server to allow AI-agent queries and traceability.

• Web3 Stack: HardHat & Foundry for development and testing, and contracts support sponsor integrations with protocols like Filecoin, Flow, and V-layer.

Notable hacks:

• We parse cross-referenced timeline events in GitHub — which are automatically created when a pull request mentions an issue — to detect which PRs are meant to fix which issues, even if the contributor doesn't manually tag them in GitHub to infer PR-to-issue links without requiring contributors to explicitly tag them.

• No wallet connection required: contributors get paid based on simply providing a wallet address in text format, and providing directly on PRs (or in PR comments, when the agent asks for it). The important thing is that there’s proof that they control their GitHub identity, which is the entity that submitted the PR and eligible for the bounty reward.