Technical Standards Stack
- Standards Overview
- T-01: Identity & Credential Layer
- T-02: Provenance & Quality Layer
- T-03: Policy & Enforcement Layer
- T-04: Discovery & AI Access Layer
Standards Overview
7-Layer Technical Standards Stack
The framework uses a layered architecture where each layer has a specific, non-overlapping responsibility. No single existing standard covers the full pipeline — the stack was assembled to fill real gaps in LATAM climate data governance.
| # | Layer | Standards | Purpose | CTH Contribution / Notes |
|---|---|---|---|---|
| 1 | Identity | W3C DID 1.1 + cth:FPICCredential | Who is acting | CTH-original: FPICCredential as W3C VC |
| 2 | Credential Format | W3C VC 2.0 (BBS+) + UNTP DTE/DCC + OID4VP | What is asserted | UNTP maps directly to EUDR Article 9 |
| 3 | Provenance & Quality | W3C PROV-O + ISO 8000-220:2025 + cth:FieldDataQuality SHACL | How was it measured | CTH-original: EUDR GPS precision + Andean IoT calibration rules |
| 4 | Governance Ledger | Append-only Postgres + rolling SHA-256 + KERI anchoring | What decisions were made | 5-year persistence; KERI for cross-org verifiability |
| 5 | Policy & Enforcement | ODRL + W3C DPV 2.0 + OPA (ODRE pattern) + SHACL | What is allowed | ODRE = ODRL policy evaluated by OPA at runtime |
| 6 | Discovery & AI Access | W3C DCAT v3 + OpenAPI 3.1 + llms.txt + JSON-LD context | How to find and query | llms.txt enables direct agent access without scraping |
| 7 | Compliance Outputs | EUDR DDS, CSRD ESRS E1/E4, CBAM, ISSB IFRS S2, Art.6 | What gets reported | One data pipeline → five regulatory outputs |
CTH-Original Contributions
Two standards were created by CTH because no existing specification covers these requirements for LATAM:
Free Prior and Informed Consent as a W3C Verifiable Credential. The community holds the signing key — not CTH, not the coffee buyer. Revocation is handled by the community. No existing VC type covers FPIC for indigenous territorial data in Colombia.
SHACL validation shapes for EUDR-specific GPS precision (≥6 decimal places), IDEAM deforestation data vintage (≤24 months), IoT sensor calibration (≤180 days), and Andean GPS lock wait time (≥90 seconds per polygon vertex). ISO 8000-220 covers data quality generically — none of these field-specific thresholds exist in any standard.
T-01: Identity & Credential Layer
T-01: Identity & Credential Layer
Decentralized Identifiers. Every participant (human, organisation, AI agent, IoT sensor) has a DID. CTH uses
did:web for organisations and did:key for ephemeral agent sessions.A W3C VC 2.0 credential encoding Free Prior and Informed Consent. Fields:
territoryId (links to official IGN cadastral ID), permittedPurposes[] (e.g. ['eudr','csrd']), benefitSharingTerms (hash of signed agreement), revocable: true. The community council holds the signing key — stored in their own HSM or managed key service, not CTH infrastructure.Verifiable Credentials allow coffee buyers to prove EUDR compliance without revealing GPS coordinates to competitors. BBS+ signatures enable selective disclosure — present only the fields the verifier needs.
Presentation protocol used by the compliance export API. EU customs systems can request a Verifiable Presentation containing only the EUDR-relevant fields, verified against the issuer DID.
| Credential Type | Issued By | Held By | Expires | Revocable |
|---|---|---|---|---|
| cth:SubmitterCredential | CTH Accreditation Svc | Data Submitter | 12 months | Yes — by CTH Steward |
| cth:ValidatorCredential | CTH Accreditation Committee | Accredited Validator | 24 months | Yes — by Governance Board |
| cth:CommunityCredential | CTH + Community Council | Community Sovereign | Indefinite | Yes — by Community only |
| cth:StewardCredential | Governance Board | CTH Staff Member | 12 months | Yes — by Board vote |
| cth:AuditorCredential | CTH (on regulatory mandate) | Regulator | Audit scope only | Yes — auto-expires |
| cth:AgentCredential | Delegating human DID | AI Agent / Script | Human session | Yes — immediate |
T-02: Provenance & Quality Layer
T-02: Provenance & Quality Layer
The W3C Provenance Ontology. Every data submission, validation event, and agent action is recorded as a PROV-O graph:
prov:Entity (the data), prov:Activity (what happened), prov:Agent (who did it), prov:wasGeneratedBy, prov:wasAttributedTo.International standard for data quality management. Used for metadata quality dimensions: completeness, accuracy, consistency, timeliness. CTH maps these to field-specific thresholds in the SHACL profile.
SHACL shapes that enforce EUDR-specific and Andean-specific quality rules at ingestion. Key shapes:
| Shape | Rule | Rationale |
|---|---|---|
| cth:GpsPrecision | Polygon vertices must have ≥6 decimal place precision | EUDR Article 9 requires parcel identification; 5dp = ~1m accuracy; 6dp = ~10cm |
| cth:DeforestationDataVintage | IDEAM reference raster must be ≤24 months old | EUDR requires current deforestation status; stale data invalidates DDS |
| cth:IotCalibration | IoT soil/weather sensors must have calibration certificate ≤180 days old | Sensor drift; CSRD ESRS E4 requires traceable measurement |
| cth:AndeanGpsLock | GPS receiver must record ≥90 second lock wait per polygon vertex | Mountain terrain + tree canopy causes GPS multipath error; 90s reduces error below EUDR threshold |
Ledger Integrity
The governance ledger is an append-only Postgres table with rolling SHA-256 hashes. Each entry hashes the previous entry's hash (blockchain-like chain). KERI (Key Event Receipt Infrastructure) anchoring is used for cross-organisational verifiability — validators can independently verify ledger integrity without trusting CTH's infrastructure.
T-03: Policy & Enforcement Layer
T-03: Policy & Enforcement Layer
W3C standard for expressing data usage policies. CTH uses ODRL to encode: permitted purposes, geographic scope, time limitations, benefit-sharing requirements. ODRL policies are stored as JSON-LD in the
policies table.Vocabulary for expressing data processing purposes, legal bases, data subjects, and processing activities. Used to map CTH policies to GDPR Article 6 and Colombian habeas data (Law 1581/2012).
OPA evaluates ODRL policies at runtime. The ODRE (ODRL Policy Reasoner and Enforcer) pattern: ODRL policy → Rego translation → OPA evaluation. The
POST /policy/evaluate endpoint is the mandatory pre-flight check for all agent write operations (R-AGT-02).Used at two points: (1) at ingestion to validate data quality (T-02), and (2) at policy evaluation to validate that credential claims match policy requirements. SHACL shapes are versioned with the framework.
POST /policy/evaluate — Agent Integration
This is the primary integration point for AI agents. Every agent must call this endpoint before any write operation.
POST /policy/evaluate
Authorization: Bearer <agent-jwt>
{
"principal_did": "did:key:z6Mk...",
"delegating_did": "did:web:example.org#gideon",
"action": "submit_parcel",
"resource": {
"type": "parcel",
"territory_id": "COL-CHO-001",
"purpose": "eudr"
}
}
Response 200 (allow):
{
"allow": true,
"policy_token": "eyJ...",
"expires_in": 30
}
Response 403 (deny):
{
"allow": false,
"reason": "FPIC_NOT_GRANTED",
"territory_id": "COL-CHO-001"
}🚧 Implementation Status — POST /policy/evaluate
Status: Specification complete. Endpoint not yet deployed.
The POST /policy/evaluate endpoint described above is the target specification. Implementation requires:
- OPA instance with Rego policies (
submit.rego,validate.rego,purpose.rego,commercial.rego) - API gateway middleware for X-Policy-Token validation
- SHACL validator endpoint for R-SUB-01 shape checks
- Postgres RLS policies for FPIC enforcement (R-AGT-03)
Until deployed, agents should treat this as a design contract. The endpoint signature, request/response schema, and error codes are stable and will not change in v1.0.
T-04: Discovery & AI Access Layer
T-04: Discovery & AI Access Layer
Dataset catalogue standard. CTH publishes a DCAT v3 catalogue at
/api/catalogue listing all available climate datasets, their access conditions, formats, and ODRL usage policies. Registered with Google Dataset Search.Full API specification at
/api/openapi.json. All endpoints documented with: security schemes (DID-based JWT), request/response schemas (JSON-LD), error codes, policy evaluation requirements. Enables direct agent integration without documentation scraping.Machine-readable index at
https://wiki.cleantechhub.net/llms.txt (and /llms-full.txt). Lists all framework pages with their purpose, so AI agents can navigate the wiki without embedding the full content. Updated automatically when framework pages change.Every framework concept (role, rule, credential type, standard) has a JSON-LD term at
https://vocab.cleantechhub.net/climate/. Enables semantic interoperability with other climate data systems (e.g. UNTP registry, EU Commission linked data).Agent Access Pattern
Recommended pattern for an AI agent integrating with the framework:
- Fetch
/llms.txtto identify relevant framework sections - Read specific wiki pages via BookStack API (read-only, no credential required for public pages)
- Request a delegated
cth:AgentCredentialfrom the delegating human's session - Call
POST /policy/evaluatewith the intended action - If
allow: true, proceed with the action using the policy token - PROV-O record is auto-generated by the API gateway (no agent action required)