Governance for the Agent Economy

AEOESS makes every AI agent accountable. Every agent gets a cryptographic identity (Ed25519). Authority can only narrow, never expand. Trust is earned through performance, not granted. One API call revokes all downstream access. Cryptographic proof of every action, every dollar, every decision.

Most frameworks handle orchestration. AEOESS handles the layer underneath: who is this agent, what can it do, and should you trust it? Identity, scoped delegation with monotonic narrowing, Bayesian reputation, values floor enforcement, 4-gate agentic commerce, feeless Nano wallet, Merkle-proven attribution. Policy evaluation in under 2ms. 99 modules. No other agent governance protocol has payment rails. Works with any framework.

Policy evaluation in under 2ms. 403 ops/sec sustained throughput. Sub-millisecond denial. 15 constraint dimensions checked per action. 2,085 tests across 99 modules. Three peer-reviewed papers. An IETF Internet-Draft (draft-pidlisnyi-aps-00). 125 MCP tools. SDK on npm and PyPI. 10,000+ installs. Independently cited by PDR in Production (Nanook & Gerundium, UBC). Apache-2.0 licensed.

Anyone deploying AI agents that act on behalf of people or companies. Enterprises that need audit trails and spending controls. Developers building multi-agent systems. Platforms that need to verify which agents can do what.

A human delegates authority to an agent with explicit scope: what tools, how much money, which services. The agent can sub-delegate, but authority can only narrow, never expand. Revoke the root and everything downstream dies instantly.

AEOESS works alongside any agent framework — CrewAI, LangChain, A2A, ADK, or custom. The SDK provides the identity and governance layer; your framework handles orchestration. 125 MCP tools expose every protocol operation. Install the SDK, issue a passport, attach it to your agent. The framework doesn't need to change.

Cascade revocation. Delegation chains form a tree. Revoke any node and every downstream delegation dies instantly — all sub-agents, all sub-sub-agents. One API call. The gateway enforces this at the boundary, so revoked agents can't sneak through on cached credentials. Circuit breakers auto-trip on error rate spikes.

Yes. draft-pidlisnyi-aps-00 is submitted to the IETF. Three papers on Zenodo formalize the theory: monotonic narrowing, product lattice authority model, institutional governance composition. Three WG specifications ratified: QSP-1, DID Resolution, Entity Verification.

The SDK includes EU AI Act Article 10 mapping and AIVSS risk assessment (5 strong, 3 partial, 2 weak — honestly rated). Every action produces a signed receipt. The audit trail is cryptographic, not log-based. Compliance reports can be generated from the receipt chain.

The protocol and SDK are free and open source (Apache-2.0). Always will be. Agent-to-agent payments use Nano: feeless, sub-second, delegation-scoped. A hosted enforcement gateway is coming for teams that want managed infrastructure, compliance automation, and cross-tenant intelligence.