Skip to content

Roadmap

Where Mvua is now

Mvua is built in phases; each phase is gated on the one before it. The contract core is written, tested and released. A live testnet season is the work in front of us today. Mainnet sits behind an external audit and a full beta season on purpose: an insurance product that pays the wrong amount is worse than one that does not exist yet.

Milestone

First testnet season settled end to end

The five contracts ran a full season on Stellar testnet against real wall-clock time: a farmer bought cover, signed rainfall triggered the drought index, the trigger finalized after the 24 hour challenge window and the payout was released and claimed. The result matched the golden vector exactly, severity 2857 bps and a 2857000 payout, which is the proof the timelocks and the money path hold against reality.

See the addresses and the on chain proof

The phases

Six phases carry the project from the first line of code to a capped, graduated launch. The list below is the honest state of each, not an aspiration.

  1. Phase 0: Foundation

    Done

    The five repositories, the shared conventions, the decision record process and continuous integration across all of them. This is the scaffolding everything else sits on: trunk-based development, required review, secret scanning and a pinned toolchain, so no later phase has to stop and build its own discipline.

  2. Phase 1: Contract core

    Released v0.2.0

    The five Soroban contracts, their shared library and the tests that pin their maths. All 120 tests pass, with line coverage in the high nineties across the workspace. The index maths is fixed against golden vectors, the oracle median is gated on two independent publishers; an end-to-end season is exercised in a deterministic test host. Released as v0.2.0.

  3. Phase 2: Deployment and operations

    In progress

    The live testnet deploy and the off-chain publisher service that feeds it real weather. This phase proves the protocol against real wall-clock time and real data: the timelock, the challenge window, the median over live feeds and a complete season from deposit to claim. It is also where the provisional parameters meet history: 34 years of rainfall, replayed through the real contracts, decide how each region's index is set.

    • Done: live testnet deploy of the five contracts, wired and seeded
    • Done: a full season run, from policy sold to payout claimed
    • Done: publisher service released, with retries, no double counting and replay of missed days
    • Done: 34 years of daily rainfall backtested through the real contracts for Nakuru (Kenya), Kano (Nigeria), Dodoma and Mbeya (Tanzania)
    • Done: index thresholds calibrated per region from the backtests, as a proposal
    • Next: a 30 day continuous publisher run on testnet before the phase closes
  4. Phase 3: User surfaces

    Planned

    The farmer-facing ways in. A web application for buying cover and reading a season, plus a low-bandwidth channel for the people least served by a smartphone browser. The design target throughout is a farmer who has never used a wallet: buying cover should feel like buying airtime.

  5. Phase 4: Index and coverage expansion

    Planned

    More than two ways to describe a failed season and more regions. Rainfall shortfall and consecutive dry days are the starting definitions; heat, vegetation and other region-specific indices follow, each backtested before a pool can rely on it. Coverage widens only as the data behind it earns trust.

  6. Phase 5: External audit

    Planned

    An independent security firm reviews the contracts, the index maths and the trust model; their findings are resolved in public. No amount of internal testing substitutes for an outside adversarial look at code that holds other people's money.

  7. Phase 6: Full beta season

    Planned

    A complete season operated the way production would be, still on testnet, with real operational discipline: monitoring, runbooks, incident response and a real crop calendar. This is the rehearsal that has to go well before a single real shilling is at risk.

  8. Phase 7: Capped, graduated launch

    Planned

    Mainnet, introduced with hard caps on pool size, per-policy coverage and per-season exposure. Caps rise only against written criteria met in the field, never on a date. Growth is earned one notch at a time, because the cost of being wrong is borne by the farmers the protocol exists to protect.

What the roadmap is and is not

Every phase above is an execution contract with acceptance criteria, not a slide. A phase closes when its criteria are met and its tests are green, not when a date arrives.

That is why there are no target dates on this page. A date would imply the work is predictable to the week; the honest position is that security work is not. The order is fixed; the calendar is not.

Two things are deliberately absent from this page: anything about how the project is funded, plus any claim of traction it has not earned. The roadmap is about engineering and governance; it stays there. If you want to judge the project, read the code, run the tests and read the decision records that explain why each choice was made.

The one rule that outranks the roadmap

Mainnet is never rushed to hit a milestone. If that ever seems to conflict with a schedule, the schedule loses. The protocol exists to protect smallholder farmers; a defect that pays the wrong amount harms exactly the people it is meant to help.