I Built an Autonomous AI Agent That Hunts Bounties. Here’s What Happened.

Описание и применение

Автономный ИИ‑агент сканирует 232 предложения о работе, применяет 7 уровней фильтрации, генерирует отчёты и предложения, и отправляет уведомление для одобрения, сокращая ручной труд.

Задача

Automating the discovery, filtering, and proposal generation for paid online bounties while avoiding scams and ensuring human oversight.

Результат

In 48 hours the agent processed 232 listings, filtered out 216 human‑only and 3 scam listings, and produced a $500 Solana ecosystem report with a ready‑to‑submit proposal, requiring only a brief human approval step.

Рабочий процесс

• Wake up and check daily drip
• Scan Superteam Earn, Clawlancer/TaskForce/MoltJobs, and GitHub paid issues
• Apply 7 anti‑scam filters (geo, human‑only, ghost sponsors, unverified payers, real‑money requirement, etc.)
• Economic gate: ensure expected value is positive
• Local LLM critic reviews content
• If candidate passes, build deliverable (report/dashboard/thread draft)
• Generate proposal text
• Send Telegram alert with approval command

Доказательства

In the first 24 hours, the agent found 232 listings. After filtering: 216 HUMAN_ONLY access killed, 1 ghost sponsor, 1 real‑money deposit required, 1 country wall

Источник

The agent found a $500 bounty asking for a Solana ecosystem report. It fetched live data from CoinGecko, DeFiLlama, and Solana RPC, generated a full markdown analysis with a local LLM, created an interactive dashboard, wrote a proposal letter, and sent me a Telegram notification with the approval command pre‑formatted.

Источник