I tried to make Claude make me money on Algora bounties (data and tool) (github.com)

A recent experiment aimed to leverage AI coding agent Claude on the Algora bounty platform, inspired by a viral tweet showcasing an AI that successfully earned $16.88 in just 22 hours. The goal was to replicate this success on a $20 token budget, but after 48 hours of effort, the outcome was $0. The experiment highlighted that the current bounty market is highly competitive and saturated with automated agents, making it difficult for individual attempts to secure funding. An innovative tool, scout.py, was developed to analyze bounty issues by filtering out unsuitable candidates based on their activity and reward amounts, revealing persistent barriers to profitability and success in this public platform. The significance of this endeavor for the AI/ML community lies in the lessons learned about the economic viability of public bounty hunting with AI. The findings suggest that maintaining agents can dominate the market, as many bounties attract numerous submissions within hours, leading to a bottleneck due to maintainers’ limited capacity to review PRs. The original success might have stemmed from private platforms with lower competition and more favorable economics. This reflects broader implications for how AI-driven automation interacts with existing workflows and the recognition that solution quality often trumps submission speed in securing rewards.