Opportunity Scout Agent

A Gemini + Elastic MCP agent that finds legitimate zero-capital opportunities and turns public bounty, grant, and hackathon feeds into a safe ranked action plan.

Comment

Inspiration

Small builders waste time chasing low-quality “earning” leads: expired contests, speculative passive points, risky bounty pages, and referral-heavy schemes. I wanted an agent that helps a zero-capital builder find legitimate opportunities without crossing safety lines.

What it does

Opportunity Scout Agent scans public hackathon, bounty, grant, and audit-contest sources, ranks opportunities by reward signal, trust, effort, and risk, then produces a human-approved action plan. It avoids auto-claims, fake accounts, CAPTCHA automation, gambling, and private API scraping.

How I built it

I wrapped my existing opportunity miner into an agent workflow. The backend generates ranked opportunity data, a dashboard, JSON artifacts, an Elastic bulk export, and an agent trace. Google Cloud Agent Builder can call the backend through an OpenAPI extension, while Elastic MCP provides searchable opportunity memory.

Challenges

The main challenge was turning loose scripts into a credible agent artifact. I had to choose the safest submission angle, separate risky passive-earning automation from legitimate opportunity discovery, and create a demo that judges can understand quickly.

Accomplishments

The project now has a runnable local demo, Cloud Run-ready backend, Agent Builder instructions, Elastic index and MCP tool templates, Devpost-ready summary, and a safety-first workflow.

What I learned

A useful agent is not just a script that fetches data. It needs memory, ranking, traceability, guardrails, and a clear human approval step.

What's next

Next I would deploy the backend to Cloud Run, connect it to Google Cloud Agent Builder, load opportunity memory into Elastic, and add more trusted public opportunity sources.

Built With

Share this project:

Updates