Give your AI agents provable identity, human delegation, and compliance scoring. 5 lines of code. Any framework.
langchain-gns-aip
Callback handler + delegation tool for any LangChain agent. Auto-collects breadcrumbs on every LLM call, tool use, and chain step.
openai-gns-aip
Lifecycle hooks for OpenAI Agents SDK. Wraps any agent with identity provisioning, delegation verification, and compliance middleware.
vercel-gns-aip
Edge-compatible middleware for Vercel AI SDK. Adds compliance headers to every streamText / generateText response.
crewai-gns-aip
BaseTool subclass + step/task callbacks. Every crew member gets verified identity. Manager delegates to the crew.
autogen-gns-aip
register_function tool + register_reply hooks + message hooks. One call wires identity into any ConversableAgent conversation.
gns-aip
Async httpx client with Pydantic v2 models. Foundation for all Python integrations. DelegationChain + EscalationPolicy utilities built in.
AI agents are making decisions, accessing data, and taking actions — but nobody can answer basic questions: Who authorized this agent? What jurisdiction governs it? Is it compliant?
The EU AI Act (Article 14) requires human oversight of AI systems. GDPR Article 22 demands accountability for automated decisions. FINMA requires audit trails for AI-assisted financial advice. Every regulation points to the same gap: AI agents have no identity.
GNS-AIP gives every AI agent a cryptographic identity that traces back to a human principal through delegation certificates. Agents are bound to geographic jurisdictions via H3 hexagonal cells. Every operation produces a privacy-preserving breadcrumb. Compliance scores are computed in real-time across five tiers.
Human Principal ──[delegates]──→ AI Agent ──[operates in]──→ Territory
│ │ │
Ed25519 key Ed25519 keypair H3 hex cells
Signs delegation cert Collects breadcrumbs Jurisdiction binding
│ │ │
└──── Delegation Chain ────────┴──── Compliance Score ─────┘You are not your iris. You are your trajectory.