English | 中文

Independent researcher building a five-layer architecture for verifiable autonomous systems.

I am an independent researcher working on protocol-governance-verification architecture for increasingly autonomous AI systems. The current main line is Digital Biosphere Architecture, a five-layer stack for persona, interaction, governance, execution integrity, and audit.

This work is not centered on shipping a single agent product. The focus is on durable architecture layers, protocol surfaces, runtime control, replay-verifiable integrity, and audit-ready evidence.

• digital-biosphere-architecture is the single canonical interpretive entry for the current five-layer stack.

• Short bio page

Responsible for persona portability and persona object structure. Not the governance, execution, or audit repository.

Responsible for interaction semantics across intent, action, and result objects. Not the transport, governance, or benchmark repository.

Responsible for runtime governance, policy checks, and budget-bound decision control. Not the architecture hub, benchmark suite, or audit plane.

Responsible for replay-verifiable execution integrity and runtime truth surfaces. Not the policy-governance or post-execution audit repository.

Responsible for post-execution review, verification, export, and audit control-plane outputs. Not the theory hub, benchmark suite, or runtime governance implementation.

• agent-evidence provides the semantic evidence substrate and SDK surface.
• agent-object-protocol provides adjacent interoperability and supporting protocol work.

Thin adapters and implementation-specific integrations are intentionally omitted from the front page.

• verifiable-agent-demo is the guided walkthrough surface across the stack, not the canonical theory hub or canonical runtime implementation.
• agent-governance-benchmark is the evaluation surface for scenarios and metrics, not the canonical theory hub or canonical runtime implementation.

• pFDO-Specification — historical context for earlier DPP work, not the current core stack.
• redrock-opendpp-core — prior lineage for DPP implementation work, not the current core stack.
• MCP-Legal-China — historical context for adjacent legal/tooling work, not the current core stack.
• Kinetic-Robotics-FDO-Sovereignty — historical context for sovereignty/K-RFS exploration, not the current core stack.
• AASP-Core — prior lineage repository, not the current core stack.
• ISAS-Core — prior lineage repository, not the current core stack.
• edo-architecture-index — historical index material, not the current core stack.

flowchart LR
    Persona["Persona Layer<br>POP"] --> Interaction["Interaction Layer<br>Agent Intent Protocol"]
    Interaction --> Governance["Governance Layer<br>Token Governor"]
    Governance --> Execution["Execution Integrity Layer<br>MVK"]
    Execution --> Audit["Audit Layer<br>ARO-Audit"]

Supporting evidence substrate: agent-evidence

Walkthrough demo: verifiable-agent-demo

• protocolized digital objects
• runtime governance
• replay-verifiable execution integrity
• audit-ready evidence and review

• ORCID
• Digital Biosphere Architecture
• persona-object-protocol
• agent-intent-protocol
• token-governor
• fdo-kernel-mvk
• aro-audit

• public research surface
• five-layer stack in active consolidation
• legacy repos preserved for lineage, not as primary entry points