Inspiration
Reward listings are noisy: deadlines move, issues close, and attractive payouts can be stale or misleading. RewardRadar treats every opportunity as untrusted until its source, status, eligibility, and deadline can be checked.
What it does
RewardRadar is a source-first opportunity scout. It turns a listing into an auditable result with:
- the canonical source link and retrieval time
- status and deadline evidence
- eligibility and risk notes
- a clear next action instead of an unexplained score
The same evidence can be reviewed on the web, queried through an Alexa+ style assistant, or shown as a focused Fire TV experience.
How we built it
The core is a small Python service with deterministic adapters and tests. A Next.js interface presents the evidence-first workflow and is exported to GitHub Pages. The project also includes:
- an Alexa+ MCP prototype with a Streamable HTTP-shaped contract and fixture data for reproducible demos
- a Fire TV route that turns the verified result into a ten-foot UI
- fail-closed adapters for sources that need credentials, so the demo never invents a payout or silently treats a closed item as open
Validation and honest limits
We ran the Python unit suite and production/static builds. The public demo is intentionally transparent: Alexa+ is simulated for this prototype, Fire TV is a web prototype rather than a packaged device binary, and live source integrations still require their real credentials. That makes the current behavior repeatable while keeping the path to production explicit.
Links
The source, tests, documentation, and demo media are open for inspection in the linked repository.
Log in or sign up for Devpost to join the conversation.