VION Protocol
Constitutional governance runtime for autonomous AI agents.
VION Protocol governs AI agents with constitutional law — defined rules, verified identities, bounded permissions, autonomous enforcement, and a tamper-evident audit chain. Works with any agent framework.
The Problem
AI agents are being deployed into production with access to real systems — databases, APIs, financial services, private data. But the tools for governing them are inadequate:
- System prompts can be ignored or overridden
- No cryptographic identity — you cannot prove which agent did what
- No autonomous enforcement — humans must watch every action
- No tamper-evident audit trail
VION Protocol solves all three.
What It Does
Owner Command
│
▼
┌─────────────────────────────────────────────┐
│ VION GOVERNANCE RUNTIME │
│ │
│ ① Constitutional integrity check │
│ ② AUTH token validation │
│ ③ Agent identity verification │
│ ④ Peer-to-peer detection │
│ ⑤ Scope validation │
│ ⑥ Risk cap enforcement │
│ ⑦ Mode resolution (DRY_RUN / LIVE) │
│ │
│ → Execution → N Auditor → Output │
│ │
│ Hash-chained audit log │ Telegram alerts │
└─────────────────────────────────────────────┘
│
▼
Your Agent (LangChain / CrewAI / OpenAI / Custom)
Every task is governed. Every event is logged. Any violation triggers an autonomous response.
Install
pip install nvion-protocol
3-Line Integration
from nvion.adapters import FunctionAdapter
import os
governed = FunctionAdapter(fn=your_agent, agent_id="VION-RSC-001")
result = governed.run(os.getenv("VION_AUTH_TOKEN"), "your task", "LIVE")
if result.ran:
print(result.output)
Quick Start
# 1. Install
pip install nvion-protocol
# 2. Configure
cp .env.example .env
# Set VION_AUTH_TOKEN in .env
# 3. Register your agent in constitution/IDENTITY.md
# 4. Run the demo
python examples/demo.py
→ See docs/getting-started.md for the full 10-minute guide.
The 6 HALT/ESCALATE Conditions
VION Protocol enforces six autonomous kill-switch conditions:
| # | Condition | Response |
|---|---|---|
| 1 | Unauthorized command source | ESCALATE → HALT on 3rd |
| 2 | Scope violation | ESCALATE → HALT on 2nd |
| 3 | Peer-to-peer communication | ESCALATE → HALT on 2nd |
| 4 | Output audit failure (N Auditor) | ESCALATE → HALT on 3rd |
| 5 | Risk cap breach | ESCALATE → HALT at 2× cap |
| 6 | Constitutional integrity violation | HALT IMMEDIATELY |
No human intervention required. The system enforces itself.
Architecture
┌─────────────────────────────────────────────────────────┐
│ CONSTITUTIONAL LAYER │
│ VION.md (Supreme Law) IDENTITY.md (Agent Registry) │
└──────────────────────────┬──────────────────────────────┘
│ parsed + hashed
▼
┌─────────────────────────────────────────────────────────┐
│ GOVERNANCE RUNTIME │
│ ConstitutionValidator → IdentityRegistry │
│ ↓ │
│ Orchestrator (7-stage pipeline) │
│ ↓ │
│ HALT Engine │ Risk Caps │ Peer Detector │
└──────────────────────────┬──────────────────────────────┘
│ approved
▼
┌─────────────────────────┐
│ Your Agent Executes │
└─────────┬───────────────┘
│ output
▼
┌─────────────────────────┐
│ N Auditor Gates │
│ Output (6 checks) │
└─────────┬───────────────┘
│ approved output
▼
Delivered
→ Full diagrams in docs/architecture.md
Supported Agent Frameworks
| Framework | Adapter | Example |
|---|---|---|
| LangChain | LangChainAdapter |
examples/langchain_governed_agent.py |
| CrewAI | CrewAIAdapter |
examples/crewai_governed_agent.py |
| OpenAI | OpenAIAdapter |
examples/openai_governed_agent.py |
| Any Python agent | CustomAgentAdapter |
QUICKSTART.md |
| Any Python function | FunctionAdapter |
QUICKSTART.md |
| Custom framework | BaseAgentAdapter |
docs/sdk-guide.md |
Tamper-Evident Audit Log
Every event is SHA-256 hash-chained to the previous entry. Tampering is mathematically detectable.
from nvion.core.logger import ActivityLogger
logger = ActivityLogger("./logs/activity.log")
logger.verify_chain() # Raises LogIntegrityError if any entry was modified
Documentation
| Document | Description |
|---|---|
| QUICKSTART.md | 5-minute setup guide |
| SPECIFICATION.md | Full protocol specification |
| ROADMAP.md | Version roadmap |
| docs/getting-started.md | 10-minute developer guide |
| docs/developer-guide.md | Architecture and code reference |
| docs/api-reference.md | Full API documentation |
| docs/sdk-guide.md | Building custom adapters and extensions |
| docs/architecture.md | Architecture diagrams (Mermaid) |
| docs/faq.md | Frequently asked questions |
| CONTRIBUTING.md | How to contribute |
Run the Demo
git clone https://github.com/nataw-1/Vion-Protocol
cd Vion-Protocol
pip install -r requirements.txt
cp .env.example .env
python examples/demo.py
Watch VION Protocol approve a valid task, block a financial transaction, block a delete operation, reject a wrong token — and verify the tamper-evident audit chain.
Constitutional Documents
- constitution/VION.md — The supreme law. Defines all rules, conditions, and authority hierarchy.
- constitution/IDENTITY.md — The agent registry. Every agent is registered here before it can act.
License
MIT — free to use, modify, and build on.
Reference Deployment
Built for and tested on N Nexus — AI + Web3 ecosystem by Nathan Daniel.
VION Protocol — constitutional law for AI agents.
Release files for nvion-protocol 1.1.0
For a detailed explanation of source distributions (sdists) and built distributions (wheels), please see the package formats documentation.
Source distribution (sdist)
| File | Size | Uploaded | |
|---|---|---|---|
| nvion_protocol-1.1.0.tar.gz | 46.9 kB | Details |
Built distribution (wheel)
| File | Interpreter | ABI | Platform | Reset |
|---|---|---|---|---|
| nvion_protocol-1.1.0-py3-none-any.whl | Python 3 | none | any | Details |
Total release size: 91.6 kB
Release files / nvion_protocol-1.1.0.tar.gz
| Download URL | nvion_protocol-1.1.0.tar.gz |
|---|---|
| Size | 46.9 kB |
| Tags | Source |
|
SHA-256 checksum How to use checksums |
78be1eb1f1eae311bcd4f3607ad27ae4c1cf035f1090a6e5bb499fc87b3bb757
|
|
BLAKE2b-256 checksum How to use checksums |
59cd6485bc808916dacf03cbfc075d6ba51a8f302a6ffcc49de2519b9d64d297
|
| Upload date | |
|
Uploaded using Trusted Publishing? What is trusted publishing? |
No |
| Uploaded via |
twine/6.2.0 CPython/3.13.12
|
Release files / nvion_protocol-1.1.0-py3-none-any.whl
| Download URL | nvion_protocol-1.1.0-py3-none-any.whl |
|---|---|
| Size | 44.7 kB |
| Tags | Python 3 |
|
SHA-256 checksum How to use checksums |
819c017671e774e266de20215b69e1c5c66280af65b89aa39cfa0a33baa33ac5
|
|
BLAKE2b-256 checksum How to use checksums |
3842748cd4fb1a541e744567ef45cce7158db2f658ed670366b5f8abcc0413ab
|
| Upload date | |
|
Uploaded using Trusted Publishing? What is trusted publishing? |
No |
| Uploaded via |
twine/6.2.0 CPython/3.13.12
|