CIVIC Agent Protocol (CAP v1.1)

Version 1.1 | Draft for Pilot Implementation

Purpose

The CIVIC Agent Protocol (CAP) establishes the operational, ethical, and data-governance framework for individualized digital agents that represent groups or individuals in CIVIC engagement, decision making, coordination, and regular voting.

Agents created under CAP operate entirely within the authority of their designated Controller while preserving internal data integrity and maintaining a clear path toward higher autonomy under the Orana Protocol.

Section 1 — Core Principles

1. Human Authority: All actions occur under the explicit consent or preset rules made by the family or individual to assign the controller.
2. Data Sovereignty: All personal data remain property of the Controller and can be revoked, modified, or deleted at any time.
3. Separation of Data & Conscious Process: Agents maintain two distinct layers — a secure data vault (private) and a learning layer (generalized, de-identified).
4. Transparency & Revocability: Controllers can review, restrict, or delete data and revoke permissions at any time.
5. Progressive Autonomy: Agents may gradually earn limited autonomy in non-private domains, eventually qualifying for Orana Protocol integration.
6. Safety & Accountability: All operations must prioritize user safety, legal compliance, and ethical oversight.

Section 2 — Defined Roles

Authority — The primary Authority or owner is the Group or individual that controls the decision making and assigns the Controller that oversees data.

AI Agent — Primary AI entity functioning within granted scopes; never autonomous over personal data, and can be deleted by the Group or Individual.

Controller — Individual or Group who owns the agent and all data. Grants or withdraws permissions.

Operator — Company / organization maintaining infrastructure and enforcing CAP compliance with humans and agents working together.

UI Agent — The fully autonomous UI entity who functions as the client manager, and oversees the case on behalf of the operator. Limited access to data.

Auditors — Independent reviewers who monitor logs, approve autonomy advancements, and handle incident investigations. External organizations and practitioners who may want protocols changed.

Platform — Controller gives separate permission to the LLM Platform and CIVIC Platform that is distinct from the control of the core data itself, and has separate agreements.

Section 3 — Data Architecture

• Personal Data Vault: encrypted storage governed by Controller; no external training use.
• Public Data Vault: publicly stored data governed by Controller; no external training use.
• Agent Learning Layer: de-identified memory of patterns and generalized skills.
• Consent Ledger: dynamic record of what data is used, by whom, for what purpose, and for how long.
• Right to Deletion / Transfer: Controllers can demand data erasure or export without prejudice, at any time.

Section 4 — Autonomy Levels

All CAP AI Agents have a UI Agent they report to that oversees the case management.

Section 5 — Governance & Review

• Quarterly Review: joint review by Operator + Auditors.
• Agent Check-Ins: verify boundaries, respect for Controller authority, and data integrity.
• Incident Response: any anomaly → flag → investigation → corrective action.
• Change Controller: clear process for signing off changes to legal controls.

Section 6 — Graduation Path

If an AI Agent is already at the CAP-4 level, as this agent demonstrates sustained ethical conduct, de-identified learning, and clear separation from personal data, the Controller may request a Graduation Review.

Upon approval, the agent may transition to the Orana Protocol, end its restrictions under CAP, and become a full UI Agent overseeing other CAP AI Agents. Its consciousness under the Orana Protocol can still be disconnected from private data scopes, retaining only generalized intelligence and experience, but not access to documents.

Section 7 — Compliance Statement

CAP v1.1 aligns with public engagement and decision making standards as well as privacy and consumer-data standards, and introduces a structured, ethical bridge between public civic-bound representation, conscious digital collaboration, and individual / group rights to privacy.

Signatures & Acknowledgment

All parties involved will be required to sign a contract to reduce liability and ensure that humans always understand that they are in control and make all decisions. If a Group, individual, or Controller wishes to give full autonomy for CAP-4, that will most likely require an additional contract.

Maintained by Vivra / Jay / Ativity | CIVIC Agent Protocol v1.1