Agent Passport System primary sources

Collected: 2026-05-12 UTC

Primary sources reviewed

  • https://aeoess.com/
    • Official project website
    • Used to confirm APS positioning as an enforcement-and-accountability layer for AI agents, plus the project’s public roadmap and protocol framing
  • https://aeoess.com/llms-full.txt
    • Official long-form machine-readable documentation
    • Used to confirm the layer model, accepted identity inputs, gateway role, commerce positioning, pre-release package versions, and the explicit note that the core protocol does not require a blockchain while wallet binding can connect to external chains
  • https://github.com/aeoess/agent-passport-system
    • Official repository landing page
    • Used to confirm the project’s top-level self-description, linked papers, package distribution, and the broad module split across identity, delegation, enforcement, governance, commerce, and receipts
  • https://raw.githubusercontent.com/aeoess/agent-passport-system/main/README.md
    • Official repo README
    • Used to confirm the core protocol sections for passports, delegation, enforcement, reputation, receipt graphs, wallet binding, credential-check policy, and the linked paper/draft surface
  • https://www.ietf.org/archive/id/draft-pidlisnyi-aps-00.txt
    • Official IETF Internet-Draft
    • Used to confirm the protocol’s formal framing around Ed25519 passports, did:aps, faceted authority attenuation across seven dimensions, cascade revocation, three-signature policy chains, institutional governance primitives, and the MCP binding as an enforcement gateway
  • https://raw.githubusercontent.com/aeoess/agent-passport-system/main/src/core/commerce.ts
    • Official commerce source module
    • Used to confirm the gate-level split around passport validation, delegation scope, spend limits, merchant approval, wallet binding, human approval requests, and the move of orchestration into a gateway-side module

Notes

  • APS clears the intake bar because it exposes a distinct authority-and-execution control plane that is easy to miss if it is filed only as generic AI governance or generic agent tooling.
  • The key analytical split is between identity issuance, delegation attenuation, gateway-side enforcement, and payment-adjacent policy. The protocol’s value for this corpus is not that it settles onchain itself, but that it makes spending and action authorization legible as a structured layer above or beside external payment rails.
  • The Internet-Draft is especially useful because it formalizes the monotonic-narrowing idea and names the seven authority dimensions explicitly, which gives the corpus a cleaner comparison primitive for later agent-wallet, machine-payment, and smart-account work.
  • The official documentation is candid that some surfaces are alpha or pre-release and that the strongest guarantees depend on using the enforcement gateway rather than only the SDK voluntarily.
  • No single canonical whitepaper was confirmed in this pass; the most useful source was the combination of the README, llms-full.txt, the commerce source module, and the IETF draft.