• Strategic agentic AI implementation planning
• EU AI Act compliance automation
• Multi-authority verification setup
• Custom integration development
• Team training and standards adoption
Published ArXiv cs.MA • June 11, 2025
AgentFacts The Universal KYA Standard for Agentic AI
Like nutrition facts for food, AgentFacts delivers standardized transparency for agentic AI systems. A foundational cybersecurity infrastructure for verified autonomous AI coordination.
Real-World Adoption
AgentFacts is actively implemented at Universitas AI for digital twin creation and recognized by leading research institutions. Enterprise pilots underway in financial services, gaming, and Web3 organizations.
Know Your Agent (KYA)
Just as consumers need standardized information about food before purchase, organizations need standardized metadata about AI agents before deployment. AgentFacts provides the transparency and verification infrastructure necessary for confident enterprise adoption.
Nutrition Facts
Serving Size 1 cup (240ml)
Calories
150
Total Fat
8g
Sodium
470mg
Total Carbs
15g
Protein
5g
Know what's in your food
Agent Facts
Know what's in your agentic AI
1. Core Identity: Research Assistant v1.0
2. Baseline Model: Claude-3-Sonnet
3. Classification: Assistant/Enterprise
4. Capabilities: Analysis, Research
5. Authentication: OAuth2 Required
6. Compliance: EU AI Act Limited Risk
7. Performance: 99.5% Availability
8. Supply Chain: Anthropic Provider
9. Verification: Multi-authority signed
10. Extensibility: Custom domains
Know what's in your agentic AI
Universal 10-Category Framework
Ten comprehensive categories providing standardized metadata for agentic AI discovery, verification, and governance. Supports JSON and JSON-LD formats for maximum interoperability.
1
Core Identity
Unique identification, names, versions, TTL management
2
Baseline Model
Foundation model transparency and safety evaluations
3
Classification
Agent type, operational level, stakeholder context
4
Capabilities
Tool calling, domain expertise, data formats, APIs
5
Authentication & Permissions
Time-limited access control with audit trails
6
Compliance & Regulatory
EU AI Act, NIST, GDPR, sector-specific standards
7
Performance & Reputation
Latency, availability SLAs, accuracy scores
8
Supply Chain
SBOM integration, dependencies, infrastructure
9
Verification
Cryptographic signatures, trust policies
10
Extensibility
Custom domains, hooks, backward compatibility
Schema Example
Real implementation following the 10-category structure (abbreviated for clarity):
JSON
{
"core_identity": {
"agent_id": "did:agent:openai:gpt-4o-mini-assistant-v2",
"name": "GPT-4o Mini Assistant",
"version": "2.0",
"created": "2024-12-01T10:00:00Z",
"ttl": 86400
},
"baseline_model": {
"foundation_model": "gpt-4o-mini",
"model_provider": "OpenAI",
"model_version": "2024-07-18",
"safety_evaluation": "openai-moderation-v1"
},
"classification": {
"agent_type": "assistant",
"operational_level": "conversational",
"stakeholder_context": "enterprise"
},
"capabilities": {
"tool_calling": true,
"supported_formats": ["text", "json", "code"],
"domain_expertise": ["general_knowledge", "programming", "analysis"],
"max_tokens": 16384
},
"authentication_permissions": {
"auth_method": "api_key",
"permissions": ["read", "generate"],
"rate_limits": {
"requests_per_minute": 3000,
"tokens_per_minute": 200000
}
},
"compliance_regulatory": {
"eu_ai_act": {
"risk_level": "limited",
"transparency_obligations": true
},
"gdpr_compliant": true,
"data_retention": "30_days"
},
"performance_reputation": {
"availability_sla": 99.9,
"avg_response_time_ms": 850,
"accuracy_score": 0.94,
"last_evaluation": "2024-11-15T00:00:00Z"
},
"supply_chain": {
"infrastructure": "Azure OpenAI",
"dependencies": [
{"name": "tiktoken", "version": "0.5.1"}
],
"data_centers": ["us-east", "eu-west"]
},
"verification": {
"signatures": [{
"authority": "openai_official",
"timestamp": "2024-12-01T10:00:00Z",
"signature": "sha256:a1b2c3..."
}],
"trust_score": 0.98
},
"extensibility": {
"custom_domains": ["enterprise_config"],
"hooks": {
"pre_request": true,
"post_response": true
},
"backward_compatible": true
}
}Industry Applications
Digital Twins
Agent verification for higher education and professional credentialing applications
Gaming & Content
Platform certification and content rating standards for agentic AI environments
Enterprise Governance
Risk management and audit trails for autonomous AI deployment at scale
Healthcare & Safety
Patient safety and regulatory oversight for medical agentic AI applications
Getting Started
1. Generate Your First AgentFacts
Use our AgentFacts generator to create standardized metadata for your AI agent. Input basic information and get validated JSON output immediately.
Open Generator2. Integrate with MCP Server
Deploy the AgentFacts MCP server for seamless integration with Claude Desktop and other MCP-compatible applications.
View MCP Setup →3. Enable Multi-Authority Verification
Configure cryptographic signatures from trusted authorities to validate agent capabilities and compliance status.
Learn Verification →Enterprise Deployment
AgentFacts enables systematic agent onboarding, governance, and compliance automation. The specification supports complex permission hierarchies, multi-jurisdictional compliance, and supply chain transparency required by enterprise organizations.
Employee Agent Pattern
Integrate third-party agents as workforce members with verified capabilities, dynamic permissions, and performance tracking.
- • Identity establishment through verified metadata
- • Role assignment with time-limited permissions
- • Compliance automation and audit trails
- • Supply chain transparency and risk assessment
- • Performance monitoring and contract management
Research & Resources
ArXiv Publication
Full technical specification published in ArXiv cs.MA with comprehensive methodology documentation.
Read Paper →GitHub Repository
Open-source implementation under Apache 2.0 with specification, tools, and examples for developers.
View Repository →Expert Implementation Services
Jared James Grogan, creator of the AgentFacts standard, provides specialized consulting for enterprise agentic AI deployment, compliance automation, and standards integration.
Contact: jared.grogan@post.harvard.edu