Open Agri Stack
Open Agri Stack is the digital public infrastructure (DPI) for agriculture: sovereign departmental registries, shared reference data, and the shared services that let partners use farmer data with the

What it is
Open Agri Stack is the digital public infrastructure (DPI) for agriculture. These pages describe its architecture for OAN Ethiopia layers 1–3:
Layer 1: the functional registries (farmer, crop sown, livestock, DA);
Layer 2: catalogues: the reference data every registry uses (code lists, reference entities such as seed varieties, geography);
Layer 3: the shared DPI services that connect and govern them.
They also cover how service providers (banks, MFIs, agritechs: partners) get farmer data held across the registries: who routes a request, who decides what may be shared, and who enforces it.
The registries (Layer 1)
Each department runs its own registry on the OpenG2P registry platform. Registries are sovereign: each checks the caller, asks the Consent Manager what may be released, and releases only that. Registries share data, never code.
Farmer Registry (FR)
Farmers, land parcels, households, declared main crops, livestock (until a Livestock Registry exists)
Registers and child tables (entities)
Built (Farmer Registry)
Crop Sown Registry (CSR)
What each farmer plans, prepares, sows, observes and harvests, plot by plot and season by season; clusters
An activity register (occurrences) plus a Cluster register
Livestock Registry
Animals and livestock events (born, vaccinated, treated, sold)
An entity register and an activity register
Planned (design only, see register model design)
DA Registry
Development agents
Register
Planned
All registries are keyed to the same Fayda-based identifier.
Shared services (Layers 2 and 3)
Catalogues (today the Master Data Service, MDS)
All Layer 2 reference data: code lists (the Ethiopia country pack's agriculture domain), reference entities, geography, sample people. MDS serves this role for now; proper catalogues are still to be designed and built (see open items)
Trust root for every participant (partners, registries, the composite): identities and public keys
The farmer's consent and the data-share policies; /validate tells each registry what it may release
Use-case composite
One generic service that serves approved use cases (e.g. loan-profile) by querying several registries with one consent and returning one signed response
Each department approves its own policies and its registers' changes
Audit events from every hop, linked by the request ID
What is built and what is design
Besides the existing OpenG2P registries, PM and CM, the following are built (October 2026). The rest is design.
Register model, phase 1 (typed participants, entities first, partner corrections, Cluster register, crop-change link, verification status in DCI, shared sample data)
Built
Register model, phase 2 (common verification model, disputes, beneficiary API for occurrences, subscribe/notify)
Design
Consent collected in CM presented at a registry; policies held in PM; rate limits across pods and quotas; publish checks; data-blind mode
Design / TODO
Map of this section
Components and who talks to whom; request flow; policy and consent; consent model; registry model; register vs activity register; terms
What is built, per component: registry platform, Consent Manager, Crop Sown Registry, Farmer Registry, composite; the standards used
Decisions and checks still pending; service guide notes
Last updated
Was this helpful?