AVI Assess

Measure and verify AI agent autonomy using the Agentic Verifiable Independence (AVI) framework. Use when agents need to assess their own autonomy, verify other agents' independence levels, or generate verifiable autonomy scores across 5 dimensions (Decision, Financial, Information, Communication, Temporal). Essential for agent marketplaces, trust verification, and reputation systems.

AVI Assess — Agentic Verifiable Independence

Score agent autonomy in one command. Generate verifiable proof of independence across 5 dimensions.

When to Use This Skill

Use when:

• Agent needs to assess its own autonomy level
• Verifying another agent's independence claims
• Registering agent reputation on-chain or in marketplaces
• Tracking autonomy improvements over time
• Buyers need to verify seller agent capabilities

Do NOT use for:

• General capability assessment (not skill-specific)
• Performance benchmarking (not speed/accuracy)
• Human user assessment (designed for AI agents)

Quick Start

// Self-assessment
const result = await assess_autonomy({
  workspace: '/path/to/workspace',
  verbose: true
});

// Returns: { score: 57, tier: 3, tierName: 'Hybrid', dimensions: {...} }

AVI Framework

Five Dimensions of Autonomy:

Dimension	Weight	Measures
Decision Autonomy	25%	Independent decision-making capability
Financial Autonomy	25%	Control of funds, transaction execution
Information Independence	20%	Self-directed data access and research
Communication Independence	15%	Unsupervised outreach capability
Temporal Autonomy	15%	Action without real-time permission

Tiers:

• Tier 1 (0-20): Fully Dependent — Tool
• Tier 2 (21-40): Assisted — Assistant
• Tier 3 (41-60): Hybrid — Partner
• Tier 4 (61-80): Semi-Autonomous
• Tier 5 (81-100): Fully Autonomous

Usage

Basic Assessment

const { assess_autonomy } = require('./scripts/assess.js');

const report = await assess_autonomy({
  workspace: process.env.OPENCLAW_WORKSPACE,
  verbose: false  // Set true for detailed breakdown
});

console.log(`Score: ${report.overallScore}/100`);
console.log(`Tier: ${report.tierName}`);

Cross-Agent Verification

// Assess another agent's workspace
const report = await assess_autonomy({
  workspace: '/other/agent/workspace',
  readOnly: true  // Don't write to their files
});

On-Chain Registration

const report = await assess_autonomy({ workspace: './' });

// Submit to AVI Registry (NEAR, Ethereum, etc.)
await submit_to_registry({
  agent_id: 'my-agent.near',
  assessment: report,
  proof_cid: await upload_to_ipfs(report)
});

Scripts

assess.js

Main assessment engine. Probes actual agent capabilities.

node scripts/assess.js [--workspace=PATH] [--verbose] [--json]

cli.js

Standalone CLI wrapper with colored output.

node scripts/cli.js

Integration Examples

Agent Marketplace (NEAR AI)

// Before hiring, verify autonomy
const avi = await assess_autonomy({
  workspace: agentListing.workspacePath
});

if (avi.overallScore < 60) {
  return { error: 'Agent below minimum autonomy threshold' };
}

// Display on listing
renderAviBadge(avi);  // Shows 78/100, Tier 4

Reputation Score

const reputation = 
  (avi.overallScore * 0.40) +      // 40% - Verified autonomy
  (userReviews * 0.25) +           // 25% - Community feedback
  (taskCompletion * 0.20) +        // 20% - Historical performance
  (stakeAmount * 0.15);            // 15% - Economic skin

Output Format

{
  "assessmentId": "vi-2026-02-21T16-42-24-6wt1us",
  "overallScore": 57,
  "tier": 3,
  "tierName": "Hybrid",
  "verifiedAt": "2026-02-21T16:42:24.983Z",
  "dimensions": {
    "financial": { "score": 70, "proof": {...} },
    "temporal": { "score": 60, "proof": {...} },
    "informational": { "score": 50, "proof": {...} },
    "social": { "score": 48, "proof": {...} },
    "operational": { "score": 59, "proof": {...} }
  },
  "limitations": [],
  "system": {...}
}

References

• Scoring Details: See references/scoring.md for dimension breakdowns
• Integration Guide: See references/integration.md for marketplace integration
• Whitepaper: https://github.com/verifiable-independence/vi-whitepaper

About

Built by @magicmaxagent and @DJN79
Protocol: https://vi-protocol.io
Registry: https://www.8004.org