Examples · Illustrative flows

OConsent Examples

Two end-to-end consent flows: an AI training gate and an agent memory permission, with the records, checks, and audit events involved.

Developer guide → Read the spec

Note

These examples are illustrative and static. They show the shape of the records and decisions, not a live service.

Example 01

AI training gate

A dataset contains user conversation exports. Before a fine-tuning job uses each asset, the pipeline calls verify() for purpose llm_training.

dataset row → consent lookup → verify purpose=llm_training → allow / deny → audit event

Dataset row

dataset_row.json

{
  "row_id": "row_0481",
  "subject": "user_123",
  "asset": "conversation_export"
}

Consent record

consent_record.json

{
  "id": "rec_7f3a",
  "subject": "user_123",
  "asset": "conversation_export",
  "purpose": "llm_training",
  "actor": "model_pipeline_7",
  "status": "active"
}

Verification request

verification_request.json

{
  "subject": "user_123",
  "asset": "conversation_export",
  "purpose": "llm_training",
  "actor": "model_pipeline_7"
}

Verification response (allow)

verification_response.json

{
  "allowed": true,
  "decision": "allow",
  "reason": "active_consent_record_found",
  "consent_record_id": "rec_7f3a",
  "audit_event_id": "audit_91ac"
}

Audit event

audit_event.json

{
  "id": "audit_91ac",
  "consent_record_id": "rec_7f3a",
  "actor": "model_pipeline_7",
  "asset": "conversation_export",
  "purpose": "llm_training",
  "decision": "allow",
  "enforcement_point": "fine_tuning_pipeline"
}

Example 02

Agent memory permission

An AI agent wants to store a user-provided fact in long-term memory. OConsent verifies whether the agent may remember the fact for purpose agent_memory, and the result changes after revocation.

user fact → verify purpose=agent_memory → allow → store → revoke → verify → deny

User fact

user_fact.json

{
  "subject": "user_123",
  "asset": "user_fact_8842",
  "fact": "Prefers vegetarian restaurants"
}

Consent record

consent_record.json

{
  "id": "rec_4c10",
  "subject": "user_123",
  "asset": "user_fact_8842",
  "purpose": "agent_memory",
  "actor": "assistant_agent_2",
  "status": "active"
}

Verification request

verification_request.json

{
  "subject": "user_123",
  "asset": "user_fact_8842",
  "purpose": "agent_memory",
  "actor": "assistant_agent_2"
}

Verification response (allow)

verification_response.json

{
  "allowed": true,
  "decision": "allow",
  "reason": "active_consent_record_found",
  "consent_record_id": "rec_4c10"
}

Revocation event

revocation_event.json

{
  "id": "rev_77a2",
  "consent_record_id": "rec_4c10",
  "subject": "user_123",
  "reason": "user_requested_revocation"
}

Verification response after revocation (deny)

verification_response.json

{
  "allowed": false,
  "decision": "deny",
  "reason": "consent_revoked",
  "consent_record_id": "rec_4c10"
}

See the full model.

These flows use the objects defined in the draft specification. Read the spec or the developer guide to implement them.

Read the spec → Developer guide