The problem. When a freelancer, creator, or small business gets paid from another country, the price they are shown is almost never the price they pay. The real cost hides in three places: a percentage fee, a fixed fee, and — the silent one — an exchange rate quietly marked up below the mid-market rate. On a $1,000 payment from the US to India, the recipient can receive anywhere from ₹88,123 down to ₹82,227 depending on the rail chosen — a 7.3% swing — while every provider truthfully advertises a "low fee," because most of the difference is not a fee at all. It is the markup, and it appears on no invoice.

The solution. FeeLens puts all three costs under one lens. For a single payment it itemizes every slice each of six providers takes (percentage fee, fixed fee, the FX-markup cost as its own line), ranks the routes by what the recipient actually receives, and prints the hidden-cost gap: how much more the priciest route quietly keeps than the cheapest one — ₹5,895.98 on that $1,000 payment.

Built with honesty as a feature. The engine guarantees recipientGets + totalCost = amount × mid-market rate for every quote; the test suite (8 tests, all passing) asserts that conservation invariant across every provider and corridor, plus hand-computed cases, so no cost can be invented or lost in the math. Fee schedules are illustrative published values collected 2026-10-28 with per-provider sources and disclaimers in data/providers.json; exchange rates are labeled sample mid-market rates. The tool is deterministic — no AI at runtime — because a cost calculator should give the same answer every time.

How it works. Zero-dependency Node.js CLI: node src/cli.js compare 1000 USD INR ranks all routes; node src/cli.js quote 1000 USD INR paypal itemizes one route; npm test runs the suite. Setup instructions are in the README.

Build disclosure. This project was built by AI agents (Tree / Muse fleet) directed by Rene Turcios of CloudCorpRecords, who selected the problem, reviewed the fee model, and owns this entry. Disclosed here and in the repo README, in line with the hackathon's instruction to acknowledge tools used.

Working prototype: Public repo with a runnable CLI + captured real output (examples/demo-output.txt) + demo video: https://github.com/CloudCorpRecords/feelens

Source code: https://github.com/CloudCorpRecords/feelens

Technology stack: Node.js (zero runtime dependencies), node:test + node:assert, plain JSON data files for provider fee schedules and rates, ffmpeg/PIL for the demo video rendering.

Setup instructions: Node.js 18+, no dependencies, no network: clone, node src/cli.js demo, npm test.

Team details: Rene Turcios (CloudCorpRecords) — solo entrant. Built with his AI agent fleet, disclosed above.