English | 中文

This repository is the canonical architecture hub for the Digital Biosphere Architecture.

Use it for the system view, repository boundaries, and cross-repo navigation. It is the architecture entry, not the implementation entry.

• Architecture -> digital-biosphere-architecture
• Evidence -> agent-evidence
• Demo -> verifiable-agent-demo
• Audit -> aro-audit

• not the concrete evidence implementation
• not the walkthrough demo
• not the audit control plane

Digital Biosphere Architecture

Layers:

1. Persona Layer Implementation: POP (persona-object-protocol)
2. Interaction Layer Implementation: AIP
3. Governance Layer Implementation: Token Governor
4. Execution Integrity Layer Implementation: MVK
5. Audit Evidence Layer Implementation: ARO-Audit

The five-layer stack explains how autonomous AI systems can be read through:

• identity
• interaction
• governance
• execution integrity
• audit evidence

flowchart TB
    DBA["Digital Biosphere Architecture"]

    Persona["Persona Layer<br>POP / persona-object-protocol"]
    Interaction["Interaction Layer<br>AIP"]
    Governance["Governance Layer<br>Token Governor"]
    Execution["Execution Integrity Layer<br>MVK"]
    Audit["Audit Evidence Layer<br>ARO-Audit"]

    DBA --> Persona --> Interaction --> Governance --> Execution --> Audit

1. Persona Layer -> persona-object-protocol
2. Interaction Layer -> agent-intent-protocol
3. Governance Layer -> token-governor
4. Execution Integrity Layer -> fdo-kernel-mvk
5. Audit Evidence Layer -> aro-audit

agent-evidence is the concrete execution-evidence entry for packaging runtime execution into offline-verifiable evidence bundles across the execution-to-audit handoff.

See docs/fdo-profile-proposal-entry.md

See docs/fdo-terminology-alignment.md

See docs/minimal-fdo-crosswalk.md

See docs/execution-evidence-layer-position.md

See docs/vienna-poster-talking-points.md

• 60-Second Triple Script
• 3-Minute Triple Script
• Rapid Q&A Triple

https://github.com/joy7758/persona-object-protocol

https://github.com/joy7758/token-governor

https://github.com/joy7758/fdo-kernel-mvk

https://github.com/joy7758/agent-evidence

https://github.com/joy7758/aro-audit

• Canonical terms for architecture vocabulary and repository-name distinctions.
• Repository map for repository roles, layer placement, and entry guidance.
• Entry Surface Matrix for the canonical hub, evidence, demo, audit, and layer-repo surfaces.
• Interaction Layer for the interaction-layer rationale and boundaries.
• Five-Layer Stack for the one-page stack summary.
• Canonical Stack, Repository Roles, and Layer Notes in this README for the main architecture overview.
• Agent Evidence for the concrete execution-evidence entry.
• Verifiable Agent Demo for the shortest guided walkthrough across layers.

• Persona alone is not enough because identity does not declare task semantics or runtime requests.
• Interaction must remain separate because intent, action, and result objects should not be collapsed into persona objects or audit receipts.
• Governance must precede execution claims because runtime boundaries need to be explicit before behavior can be reviewed.
• Execution integrity matters because runtime claims need checkable evidence about what actually happened.
• Audit evidence must be exportable because outside reviewers need portable artifacts instead of internal-only assertions.

• Agent Evidence for the concrete execution-evidence package and offline verification path.
• Verifiable Agent Demo for a compact end-to-end walkthrough.
• POP quick trial for a minimal persona-layer entry point.
• Agent Intent Protocol for the interaction-layer draft.
• Token Governor for runtime governance and budget-policy control.
• ARO-Audit quick demo for evidence-oriented outputs.

Portable persona objects and identity attachment across tools, runtimes, and interactions.

Machine-readable intent, action, and result objects for agent-to-agent and agent-to-runtime coordination.

Policy, permissions, budget, and runtime constraints that determine what a system is allowed to do.

Runtime truth, integrity checks, and minimal verification surfaces for execution claims. MVK belongs here as execution integrity work, not as a separate parallel architecture axis.

Evidence bundles, exportable records, and reviewable audit artifacts for governance and conformance workflows.

• Threat model
• Conformance vectors
• Shortest validation loop
• Layer positioning

Download: docs/governance_architecture_whitepaper.pdf

Suggested citation:

joy7758. Governance Architecture for Autonomous AI Agents: Digital Biosphere Architecture Whitepaper. GitHub, 2026.

• Architecture
• Interaction Layer
• Five-Layer Stack
• Governance Architecture Whitepaper
• Community Discussion Summary
• Ecosystem Map
• Architecture Map
• Core Objects
• Runtime Positioning
• Standards Hardening Roadmap
• One-Page Brief

• narrative hub for the canonical five-layer architecture stack
• working draft
• ecosystem alignment in progress