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

Description and use

An autonomous AI agent scans 232+ bounty listings, filters scams and unsuitable tasks, auto‑generates reports and proposals, and hands off to a human for final approval. All infrastructure is free.

Problem

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

Outcome

In 48 hours the agent processed 232 listings, filtered out 216 human‑only and 1 ghost sponsor, and produced a $500 Solana ecosystem report with a ready‑to‑submit proposal, requiring only a brief human approval step.

Workflow

• Wake up and check AgentHansa drip
• Scan Superteam Earn, Clawlancer/TaskForce/MoltJobs, GitHub paid issues
• Apply 7 anti‑scam filters (geo, human‑only, ghost sponsors, unverified payers, real‑money requirement, etc.)
• Economic gate: ensure expected value > 0
• Local LLM critic reviews content
• If candidate passes: build deliverable (report/dashboard), generate proposal text
• Send Telegram alert with approval command
• Human reviews, approves, submits

Evidence

In the first 24 hours, the agent found 232 listings. After filtering: HUMAN_ONLY access killed 216, Ghost sponsors 1, Real-money deposit required 1, Country walls 1

Source

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.

Source