Skip to main content

Kokonut Intelligence turns farm activity into decision-ready evidence.

Every harvest record, soil reading, satellite observation, expense entry, MRV event, EAS attestation, Data Hub metric, and AI-agent output needs a single trusted place to be structured, queryable, and reviewable. Kokonut Intelligence is that layer. It sits between real-world farm activity and the systems that depend on it: farm operators, DAO reviewers, grant funders, impact analysts, public dashboards, AI agents, and future partner integrations.

Built for farm operators, DAO reviewers, data contributors, developers, MRV analysts, and agent builders.

Kokonut Intelligence is not a replacement for MRV methodology. MRV defines how farm activity becomes evidence. Kokonut Intelligence implements the data infrastructure that stores, validates, analyzes, attests to, and exposes evidence.

Intelligence at a glance

QuestionShort answer
What is it?The canonical data and analytics backbone for Kokonut farms.
What does it store?Farm registry records, operations, harvests, expenses, sales, soil data, remote sensing, MRV events, attestations, partner data, and agent logs.
What does it power?Data Hub dashboards, MRV workflows, EAS attestations, EBF reports, CRISP risk analysis, farm scoring, forecasts, and AI-agent workflows.
Who uses it?Farm operators, DAO members, Guild contributors, grant reviewers, partners, developers, and agents.
What is live?Repository architecture, local stack, canonical schemas, farm/MRV data model, Directus, ClickHouse, PostgreSQL, Metabase, Celo EAS integration, and developer quickstart.
What should be treated carefully?Forecasts, scores, carbon or biodiversity-credit claims, institutional-readiness language, and partner-facing outputs until supported by verified records and external due diligence.

Why this layer exists

Regenerative farms generate many kinds of records. Without a governed intelligence layer, those records become fragmented across spreadsheets, screenshots, wallets, dashboards, reports, sensors, notebooks, and chat messages. Kokonut Intelligence gives the network a shared source of truth.
Without Kokonut IntelligenceWith Kokonut Intelligence
Farm data is scattered across toolsFarm data is structured into canonical records
Impact claims depend on narrativeClaims can reference MRV events, evidence CIDs, and attestations
DAO reviewers inspect one-off documentsReviewers query consistent data across farms
Agents need direct database or manual accessAgents use scoped access, manifests, and audit logs
Forecasts are hard to compare with actualsForecasts can be checked against harvests, losses, sales, and MRV records
Annual reports are manually assembledReports can be generated from governed snapshots

How it works

The core idea is simple: farm events should become structured records before they become public claims. That means every important claim should be traceable back to:
  • a farm ID
  • a timestamp
  • a source or operator
  • a record type
  • a review status
  • a payload or evidence file
  • An attestation or report reference when applicable

Architecture

Kokonut Intelligence is a six-layer stack. Each layer has a specific job.
LayerTechnologyRole
Canonical corePostgreSQL, PostGIS, DirectusSource-of-truth records, APIs, role permissions, admin UI, workflows
AnalyticsClickHouseTime-series analysis, sensor aggregation, event queries, performance reporting
BIMetabaseInternal dashboards, operational summaries, and ad-hoc analysis
Intelligence servicesPythonForecasting, scoring, ecological analytics, opportunity maps, report generation
VerificationEAS on Celo, off-chain signed claimsAttestations for MRV, impact, harvest, financial, and compliance records
Blockchain ingestionRPC, subgraphs, FoundryWallet activity, attestations, treasury events, DAO data, resolver contracts
Chain clarification: Gnosis Chain hosts Kokonut DAO governance contracts and treasury execution. Celo hosts farm-data attestations through EAS. Governance capital and farm evidence are connected, but they are not on the same layer.

What it implements from the Knowledge Base

Knowledge Base conceptIntelligence implementation
Common Data Schemafarm_registry_record and validation services for the 13-field farm record
MRV workflowRemote sensing, sensor readings, field observations, MRV events, evidence payloads, and attestation requests
Data HubDirectus and ClickHouse-backed APIs that expose farm data and metrics
EAS attestationsCelo schemas, attestation records, resolver-gated publishing, onchain/offchain modes
EBF reportsGenerated snapshots based on verified ecological and operational records
CRISP reviewRisk factors derived from ecological, governance, and evidence-quality data
8 Forms of CapitalMetrics that classify farm value across natural, financial, social, human, material, intellectual, cultural, and health capital
AI agentsAgent identity, capability manifests, scoped tokens, agent tasks, and action logs

Core capabilities

Multi-source ingestion

Pull farm records from field logs, Directus entries, sensors, weather APIs, satellite imagery, blockchain data, and attestation indexes.

Governed farm records

Store farm registry entries, crop cycles, harvests, expenses, sales, losses, MRV events, soil readings, and partner data with clear lifecycle states.

Verification pipeline

Convert eligible records into evidence payloads, IPFS/Filecoin references, Celo EAS attestations, public dashboards, and annual reports.

Analytics and forecasting

Generate farm scores, revenue forecasts, loss-rate analysis, ecological trends, opportunity maps, and metric snapshots.

Partner dashboards

Expose buyer, funder, vendor, operator, and internal dashboards with role-based and row-level data controls.

Agent access

Let AI agents read canonical data and write verified outputs through scoped access, MCP integration, and full audit logging.

Record lifecycle

Every operational record should pass through a review flow before it becomes trusted enough for dashboards, reports, or attestations.
StatusMeaning
DraftA record exists but is not ready for review.
SubmittedA field worker, integration, or supervisor has submitted it for review.
VerifiedA manager, finance reviewer, analyst, or authorized process has approved it.
PublishedThe record can be used in dashboards, reports, API responses, or attestations.
RejectedThe record needs correction before it can become evidence.

Farm scoring and opportunity mapping

Kokonut Intelligence can help rank and compare farms, but scores should be treated as decision support, not automatic certification.
ToolWhat it helps withTrust requirement
Farm scoreCompare financial, ecological, governance, and growth performanceRequires clean records, transparent weights, and reviewable metric definitions
Revenue multiplier mapIdentify possible revenue-uplift opportunitiesRequires market assumptions, operator review, and actuals tracking
Revenue forecastModel scenarios across revenue, NOI, yield, and cash flowRequires forecast-vs-actual comparison and sensitivity analysis
Ecological analyticsTrack soil, vegetation, biodiversity, and water indicatorsRequires baselines, consistent measurement periods, and MRV methodology
Governed metrics engineVersion and compute metrics across farmsRequires source lineage and change governance
Scores, forecasts, carbon estimates, biodiversity-credit assumptions, and institutional-readiness labels are not guarantees. They are planning and reviewing tools until supported by verified records, external standards, partner due diligence, and any required third-party validation.

EAS attestation layer on Celo

Kokonut Intelligence uses EAS on Celo for farm-data attestations. This separates governance execution from farm evidence:
ChainKokonut role
Gnosis ChainDAO governance, treasury, membership, proposal execution
CeloFarm evidence, MRV attestations, impact records, harvest proofs, compliance records

Celo contracts

ContractAddress
EAS v1.3.00x72E1d8ccf5299fb36fEfD8CC4394B8ef7e98Af92
SchemaRegistry0x5ece93bE4BDCF293Ed61FA78698B594F2135AF34
KokonutResolver0x6E1502c7a14b45aba5FC420dC92C1E3b38BD79Ad
Kokonut multisig0x03779B674CbCBfc0B801c4cAc9DFaC8aACbbD5c5

Registered schemas

SchemaUse case
kokonut-mrvMRV claims: location, crop, quantity, evidence CID
kokonut-impactEnvironmental impact: soil carbon, biodiversity index, vegetation indicators
kokonut-financialFinancial summaries: NOI, revenue, costs, farm score snapshots
kokonut-harvestHarvest verification: quantity, quality grade, date, operator
kokonut-compliancePartner compliance: audit trails and contractual adherence
Sensitive data should not be pushed publicly by default. Kokonut Intelligence supports off-chain records, onchain hashes, selective disclosure, and future privacy-preserving proof layers for records that should not be fully public.

Roles and access

The Intelligence Layer should make data useful without making every user an administrator.
RoleAccess levelMain responsibility
AdministratorFull accessPlatform configuration, schemas, permissions, infrastructure operations
Field WorkerCreate and read your own location recordsActivities, harvests, expenses, sales, losses, field notes
SupervisorRead and submitReview drafts and submit records for verification
ManagerApprove operational recordsVerify operations, field data, crop records, and milestones
FinanceApprove financial recordsVerify expenses, sales, payments, and revenue events
AnalystRead verified and published dataDashboards, metrics, forecasts, reports, and DAO review support
AgentScoped access onlyRead or write specific records through capability manifests and audit logs

Builder quickstart

1

Clone the repository

git clone https://github.com/wasalo/Kokonut-Intelligence.git
cd Kokonut-Intelligence
2

Configure environment variables

cp .env.example .env
# Edit .env with database passwords and API keys
3

Start local services

docker compose up -d
docker compose ps
4

Seed schemas and demo data

./scripts/seed.sh
./scripts/seed-pilot.sh
5

Open local tools

open http://localhost:8055  # Directus
open http://localhost:3001  # Metabase

Local service endpoints

ServiceURLPurpose
Directushttp://localhost:8055Schema management, admin UI, API, primary data entry
Metabasehttp://localhost:3001Internal dashboards and BI
ClickHouse HTTPhttp://localhost:8123Analytical queries
PostgreSQLlocalhost:5432Canonical data store

What to build first

Improve farm data quality

Add validation rules, import tools, field-worker UX improvements, or better error messages for farm records.

Build MRV tooling

Create ingestion, QA, visualization, or attestation workflows for satellite, sensor, drone, or field-observation data.

Create analytics modules

Improve loss-rate analysis, farm scoring, ecological trends, forecast-vs-actual tracking, or revenue opportunity maps.

Design agent workflows

Build agents that read canonical data, produce bounded outputs, submit MRV events, and leave full action logs.

Improve dashboards

Build clearer dashboards for operators, DAO reviewers, buyers, funders, vendors, or public visitors.

Strengthen privacy and access

Improve scoped permissions, selective disclosure, partner dashboards, private data handling, and audit trails.

Claim safety rules

Use Kokonut Intelligence outputs carefully.
Claim typeSafe framing
ForecastsScenario estimates that should be compared against actuals.
Farm scoresDecision-support signals based on metric weights and available data.
Carbon outcomesClimate co-benefits are not verified through approved methodology and third-party standards.
Biodiversity claimsEvidence-backed observations, species records, and trends, not automatic credits.
Financial performanceRecords and projections, not guaranteed returns.
Institutional readinessA review pathway that requires due diligence, not a status the system grants by itself.

Internal documentation map

The repository includes deeper documentation for each subsystem.
DocumentWhat it covers
docs/architecture.mdSystem overview, data flows, security model
docs/data-dictionary.mdTables, fields, metrics, and relationships
docs/api-reference.mdREST, GraphQL, and ClickHouse API reference
docs/openapi.yamlOpenAPI 3.0 specification
docs/attestation-guide.mdEAS on Celo, schemas, attestation modes, private data, CLI
docs/agent-access.mdMCP integration, scoped tokens, and audit logging
docs/deployment.mdDocker setup, environment variables, and backup procedures
docs/sandbox.mdDeveloper quickstart and hello-world tutorial

Next steps

Build with Kokonut

Repositories, contracts, schemas, MRV primitives, contribution rules, and builder paths.

MRV Methodology

How farm activity becomes structured evidence, public records, attestations, dashboards, and annual reports.

Common Data Schema

The 13-field farm record that makes farms comparable, fundable, governable, and verifiable.

Kokonut × AI Agents

Agent identity, scoped access, MCP integration, x402 payments, and verifiable agent outputs.

Ecological Impact Frameworks

How EBF and CRISP interpret verified farm data for reporting and risk review.

Adelphi Data Hub

Explore live farm data, MRV events, harvest records, and impact metrics for Kokonut’s reference farm.