Representative Colombian Peso · Phase 1 · version 1.0 · data as of 29 September 2026
Academic MVP on Celo Sepolia (testnet). Not for production. The full technical white paper is in Spanish at /libro-blanco. Its section 7 contains the five binding disclaimers (D1–D5), in Spanish. They are not translated on purpose.
Issuer. rCOP is issued and was created by Grupo New Way SAS, a commercial company incorporated under the laws of the Republic of Colombia, NIT 901.855.061-7.
What it is. rCOP is an ERC-20 token that tracks one Colombian peso (COP). Users deposit USDC and receive rCOP at the published COP/USD rate; they burn rCOP to get USDC back at the same rate. There are no loans, liquidations or debt. The contract is a minimal, single-collateral fork of Angle Protocol's Transmuter on Celo. It is the applied case of a master's thesis in Law and Economics.
Mechanism.
Fees. Minting costs a constant 10 bps. The burn fee depends on the reserve ratio (RR = reserves valued in COP ÷ rCOP supply):
RR Burn fee ≥ 0.95 15 bps 0.90–0.95 50 bps 0.75–0.90 200 bps Auto-pause. Below RR 0.75, mint and burn revert until the RR recovers.
Caps. Supply is capped at 8,000,000 rCOP and minting at 800,000 rCOP per address, about USD 2,000 and USD 200 at a rate of 4,000 COP/USD.
Oracle. An off-chain bot publishes the price every five minutes. It weights three sources:
- the official TRM certified by Colombia's financial regulator (60%, and required);
- a market FX rate (25%);
- the Binance P2P USDT/COP midpoint (15%).
Sources that are stale or more than 3% away from the TRM are dropped. On-chain, the wrapper checks four things:
- the publisher role;
- the L2 sequencer status (on mainnet only);
- absolute bounds of 2,000–10,000 COP/USD;
- a 3% maximum deviation from the simple mean of the last 30 publications.
The engine also rejects prices older than 10 minutes. RedStone was the original design, but it has no COP/USD feed.
Governance. The Phase 2 target is a 3-of-5 Safe proposing to 48 h and 24 h timelocks, with an instant pauser. In Phase 1 on testnet, a single author-controlled account holds every privileged role, and governance changes take effect immediately.
Evidence and known limitations.
- Liveness. The oracle stopped publishing twice: for about 95 days (June–September 2026, when the bot ran out of gas and CI billing lapsed) and from 27 September 2026 (the external scheduler disabled the trigger).
- Structural FX mismatch. Reserves are in USD and liabilities in COP, so peso appreciation lowers the RR by itself.
- First-come redemption. Burns pay at par with no pro-rata haircut.
- Oracle manipulation. A compromised publisher key can walk the count-based mean within a single block. In tests this extracted +24.69% on 150 USDC.
- No self-recovery. After downtime the oracle can get stuck without a governance action.
A proposed oracle v3 (not deployed) adds rate limits, a time-weighted reference and self-recovery. Contracts stay frozen until the thesis defense.
Quality.
- 170 Foundry tests, including stateful invariants I1–I6.
- Per-file branch coverage gates: Engine 100%, token 100%, oracle wrapper 88.89%, ReserveMath 80%.
- Slither and Aderyn run in CI.
- An end-to-end mint/burn test through the production app matched the contract math to the wei.
- No external audit yet; it is a precondition for Phase 2.
Status of the token. rCOP is not legal tender, is not guaranteed by any public entity, pays no yield, redeems on a best-effort basis, and is not offered to the general public.