What is the fastest way to get started?

npm install agent-passport-system, then import from agent-passport-system/core for the curated essentials surface: createPassport, createDelegation, evaluateIntent, commercePreflight, generateKeyPair, plus the essentials. The full API surface is still at agent-passport-system for advanced cases. The MCP server defaults to 20 essential tools (set APS_PROFILE=full for all 150).

An open protocol for accountable AI agents

Q: What is Agent Passport System?

Agent Passport System is the enforcement and accountability layer for AI agents. Bring your own identity (did:key, did:web, SPIFFE, OAuth, did:aps). Policy eval under 2ms. 2,884 tests. 20 essential MCP tools (150 total). npm install agent-passport-system.

Q: What does the protocol do?

Makes every AI agent accountable. Cryptographic identity (Ed25519), scoped delegation with monotonic narrowing, earned reputation, values enforcement, 4-gate commerce with agent-to-agent payments, Merkle attribution, gateway enforcement boundary. Authority can only narrow, never expand.

Q: Is Agent Passport System open source?

Yes. Apache-2.0 licensed. SDK v2.6.0-alpha.1. Independently cited by PDR in Production (University of British Columbia). github.com/aeoess

Open-source enforcement infrastructure for AI agents

● Protocol LiveApache-2.0— installs

AI agents are moving money, signing contracts, deploying code, and accessing sensitive data on behalf of companies and people. Today, most companies ship without cryptographic answers to who authorized them, what scope applies, or how to revoke them downstream.

Today

Agents act anonymously. No audit trail. No spending controls. Revocation means shutting everything down. You find out after the damage.

With Agent Passport System

Every agent carries a signed identity. Authority can only narrow, never expand. One call revokes downstream. Full cryptographic audit trail.

Identity—Bring your own: did:key, SPIFFE, OAuth

Delegation—Scoped, revocable chains

Composition—Cross-DID-method identity contract

Mutual Auth—Agent ↔ service handshake

Reputation—Earned trust, scoped to context

Governance—Values floor enforcement

Commerce—4-gate spending controls

Payments—Agent-to-agent, delegation-scoped

Attribution—Merkle-proven audit trail

Instruction Provenance—Drift denial across turns

Enforcement—Gateway boundary, <2ms

Revocation—Cascade, one call kills all

<2msPolicy eval

20Essential tools

2,586Tests

Full surface area: 150 MCP tools, TypeScript + Python SDKs.

Independently cited — PDR in Production (University of British Columbia) validates APS Bayesian model. Zenodo

$ npm install agent-passport-system

$ clawhub install agent-passport-system

npm SDK v2.6.0-alpha.1 MCP Server v3.1.1 ClawHub Skill Python SDK v2.4.0a1 8 papers →

Quick Start — Core subpath, curated essentials

// import the curated essentials
import {
  createPassport, createDelegation,
  evaluateIntent, commercePreflight, generateKeyPair
} from 'agent-passport-system/core'

// full API surface still available at 'agent-passport-system'

Updates

May 03shipPhase 4.1 alpha across four registries — SDK 2.6.0-alpha.2, MCP 3.2.0, Python 2.4.0a2, Skill 5.9.0. Rail receipts as accountability evidence (claim_type + scope_of_claim), DID URI signer with rotation-aware verifier, cross-receipt link fields. 2,884 tests. npm

May 03convergenceBilateral receipt cross-impl byte-match — APS canonical-bytes path verifies 5/5 against desiorac's qntm v0.3.1 fixture, independent of the Wave 1 cross-language scenarios. Mirror offered into aeoess/aps-conformance-suite for v0.3.2 mid-May publish. qntm #15

May 02shipSDK 2.6.0-alpha.0 on npm — v2 evidentiary type safety landed: claim-evidence-types registry with BATCH_ATTESTED and EVIDENCE_CUSTODY_HELD extensions, claim-verifier module, contestation cascade with verifier hook, path-scoped cycle detection in mergeTaints. 2,884 tests passing. npm

May 02shipPython SDK 2.4.0a1 on PyPI — Full Wave 1 surface ported from TS: v2/accountability/* (action, authority-boundary, bundle, custody, contestability), v2/cognitive_attestation/*, v2/instruction_provenance/*. Cross-language byte-parity verified across 27 fixtures (15 evidentiary + 12 Wave 1). 518 tests. PyPI

May 02standardVocab phantom-issuer audit — PR #74 removed RNWY from behavioral_trust and wallet_intelligence (couldn't verify those signals are issued in production); PR #75 marked passport_grade as status: proposed (single-issuer rule). Single-source-of-truth discipline. PRs #74 + #75

May 01standardVocab PR #72 opened — completion_ratio as canonical signal type with three production issuers (AgentID, APS, RNWY). Descriptor enforcement_class advisory, completion_ratio_method constraint with strict and quality-weighted modes. PR #72

May 01shipVocab validator hardened — nested descriptor_dimensions walk under signal_types + legacy whitelist for #57 stale values. Three pre-resolution descriptor uses preserved without warning maintainers. validator

May 01shipDrift prevention infrastructure live — pre-commit hook + CI scan + standardized .gitignore + final-scan propagation across eight public repos. Four layers, one structural backstop against private-context drift.

Apr 30shipWave 1 accountability shipped — Five signed receipt primitives (action, authority-boundary, custody, contestability, bundle). RFC 8785 + Ed25519, deterministic byte-match fixtures, 57 new tests. SDK v2.6.0-alpha.1. module README

Apr 30tractionVocab PR #66 merged — Edison Munoz Duran's Agent-DID crosswalk lands as the second co-drafted-with-aeoess crosswalk. A2A composition contract co-drafting in flight on a shared spec branch. PR #66

Apr 30convergenceVeritasActa cross-layer integrity 10/10 — Knowledge Unit bundle with sidecar-anchored APS DecisionLineageReceipt verifies end-to-end. All ten access receipts cross-layer hash-matched, APS signature valid against sidecar JWKS. Tamper-detection holds across both layers. verify PR #7

Apr 29paperPaper 8: The Evidence-Safety Gap — Cryptographic agent governance proves procedural validity, not effect safety. Compliance-complete failure named, five omitted-variable classes catalogued, defeat traces in reference implementation. Zenodo.

Apr 29tractionSSRN + ORCID research surface — Three papers entering SSRN (Agent Social Contract, Physics-Enforced Delegation, Cognitive Attestation). ORCID 0009-0002-4700-3594 live with all 8 papers indexed via DOI.

Apr 29tractionVocab PR #52 merged — entity_continuity PDR validator with @nanookclaw, 309 LOC pure-Node + 32-test suite + four reference fixture vectors. Behavioral-fingerprint-drift validator complementary to continuity-analyzer's structural fixture. PR #52

Apr 29standardVocab PR #61 merged — epoch added to validity_temporal enum, observer-relative ticks on substantive state transitions distinct from sequence's event-relative counts. Issue #58 settled with @lawcontinue's endorsement. PR #61

Apr 29standardVocab PR #62 merged — governance_attestation.refusal_authority brought into formal enum compliance (structurally_impossible_to_violate → issuer). One-line correction, @lowkey-divine concurred on issue #57. PR #62

Apr 29tractionVocab PR #56 merged — AuthorPrime's Sovereign Atom crosswalk, third independent entity_continuity implementation alongside SBR-002 and continuity-analyzer. PR #56

Apr 28tractionIPR module on npm — agent-passport-system@2.4.0-alpha ships canonicalize/envelope/verify for binding agent authority to instruction-file digest at delegation. npm

Apr 28tractionVocab PR #63 merged — piiiico added trust_verify endpoint to AgentLair behavioral_trust, AAT-based lookup without agentId resolution. PR #63

Apr 27standardagent-governance-spec org created — cross-vendor spec home, co-edited with Lars Kroehl (MolTrust). spec repo

Apr 27tractionVocab PR #59 merged — DCP-AI crosswalk, composite Ed25519 + ML-DSA-65 in production. PR #59

Apr 27tractionVocab PR #53 merged — AgentNexus three-issuer fixture, Step 2 of the Interop Week 1 compose chain. PR #53

Apr 27convergenceVocab #58 epoch enum: three-way endorsement — lawcontinue, kenneives, srotzin. #58

Apr 26shipcontributor-check installed — MS AGT v3.3.0 composite action live on agent-passport-system, agent-passport-mcp, agent-governance-vocabulary. Pinned to commit 15e001f9b53f, threshold HIGH for calibration window. action source

Apr 26shipVocab PR #55 opened — APS ↔ ACTA receipt crosswalk, 14 mappings (1 exact, 7 partial, 3 divergent, 2 no_mapping, 1 non_equivalent). Calibrated against @veritasacta/* and protect-mcp. PR #55

Apr 26shipaeoess/governance-attestation-predicate live — in-toto sibling to nobulex's Decision Receipt PR #549. JWS+Ed25519, 5 fixture vectors, composition test passes. repo

Apr 26shipaeoess/aps-conformance-suite live — 37 fixture vectors across 4 categories (bilateral-delegation, inference-session, instruction-provenance, AIVSS scenarios). All byte-identical reproducible. repo

Apr 25shipVocab PR #51 opened — invariant-survival.md descriptor doc co-authored with QueBallSharken (BBIS). PR #51

Apr 25shipVocab PR #52 opened — entity_continuity PDR validator with 32 tests, co-authored with nanookclaw. PR #52

Apr 25tractionVocab PR #49 merged — PIC Standard becomes the 23rd crosswalk, modeled as parallel verification surface to visa-layer issuance. PR #49

Apr 25tractionVocab PR #46 merged — AgentLair becomes the third production issuer of behavioral_trust, alongside RNWY and Logpose. PR #46

Apr 25shipaps-system PR #19 merged — lawcontinue's seven-vector CTEF inference-session fixture pack, all RFC 8785 JCS-canonicalized and Ed25519-signed. PR #19

Apr 25convergenceharness PR #1 — MoltyCel published moltrust v0.2.0 to PyPI and opened a draft adapter PR. Mark-ready expected Monday. PR #1

Apr 24standardA2A extension proposal opened — claim_type discriminator over {identity, transport, authority, continuity}. Aligns to CTEF v0.3.1, no proto changes. issue

Apr 24convergenceFive-way claim_type convergence — AgentGraph, AgentID, APS, Nobulex, and HiveTrust aligned on the wire-format key after a mid-thread rename. #1672

Apr 24standardOpenClaw #49971 closed — steipete's ruling: trust providers build on five existing public hooks rather than a new core API. #49971

Apr 24convergenceCanonicalization loop closed with AgentGraph — three harnesses now lock through RFC 8785 JCS bytes. Five fixtures byte-identical. thread

Apr 24shipRotation-attestation fixtures v1 live at fixtures/rotation-attestation/. RFC 8785 JCS, deterministic generator. test-vectors

Apr 24shipautogen-governance-adapter PR #1 — first external security contribution. pshkv fixed a silent JWKS kid fallback in _lookup_issuer_key. PR #1

Apr 24shipHousekeeping batch shipped — seven AUDIT-2026-04-24 fixes across six packages. Audit spec bumped to v2.3. completion

Apr 23convergenceMicrosoft AGT #1354 — interop proposal posted, mapping #1386's three questions to named APS primitives. Pending response. reply

Apr 23convergenceCTEF v0.3.1 adopts APS composition-rule table verbatim — four-row per-layer grammar pulled into §6.3 as normative. thread

Apr 23shipFull Code Audit v2.1 — pre-publish spec rewritten to forty-two steps across fourteen repos. Three tiers: code integrity, supply chain, runtime. prompt

Apr 23shipPR Merge Protocol v0.2 — added Seven Deep-Review Dimensions for substance issues that pass surface checks. CONTRIBUTING.md expanded. CONTRIBUTING

Apr 23standardBBIS classification grammar adopted — Hensley's five-bucket model replaces the prior taxonomy. M4 EffectReceipt renamed to FRCBE per qntm#7. qntm#7

Apr 23shipAgent Ecosystem Directory shipped — three sortable tables: 18 projects, 115 contributors, 93 governance threads. Live GitHub data. directory

Apr 22shipautogen-governance-adapter v0.1 shipped — glue for autogen's before_tool_call hook. governedToolCall() with provider-agnostic TrustProvider, 12 tests green. repo

Apr 22tractionThird aeoess PR merged in microsoft/agent-governance-toolkit — examples/cognitive-attestation-governed/ landed via PR #1328. PR #1328

Apr 22tractioncomposed/v1 extends from three to four signals in seven hours — schchit added JEP as decision_event fourth signal. PR #8

Apr 22shipMutual Authentication v1 shipped (SDK v2.6.0-alpha.1) — downgrade-proof four-step handshake, A2A and MCP adapters, 29 new tests. module README

Apr 21shipcomposed/v1 three-signal worked examples — first AgentID + APS + AgentGraph envelope under one shared subject DID. Kenne LGTM'd. PR #7

Apr 21tractionMicrosoft AGT merged ADR 0006 on constitutional constraints, after our five posts on fanout-as-risk-signal. PR #1199

Apr 21shipa2a-compliance-harness v0.1 shipped — five-step Python check for A2A Agent Cards. repo

Apr 21shipA2A trust-header Week 3 — dual-provider verifier, 9/9 fixtures pass. MolTrust ready for drop-in. PR #6

Apr 21standardOpenLineage upstream PR opened — AgentAttributionRunFacet spec, four DCO-signed commits, CI green. First Agent Passport System contribution to LF-hosted. PR #4480

Apr 21standardcontext_dimensions added to vocabulary — third top-level section, four day-1 entries. PR #34

Apr 21tractionInterop Week 1 Step 1 fixture merged — AgentID trust_verification fixture. PR #38

Apr 20shipCognitive Attestation envelope (SDK v2.1.0) — Paper 7 primitive ported to SDK. JCS canonicalization, three-stage verification, typed dispute primitives.

Apr 20shipverifyBoundWallet object-form overload — Asymmetry with bindWallet gone. MoltyCel's ask in SDK#16.

Apr 20standardAPS filed with AAIF — Project proposal #14. Path toward Linux Foundation stewardship for the protocol layer. Commercial stack kept independent.

Apr 20shipv2.0.0 promoted to npm @latest — SDK + MCP + Python flipped from @next. v1.46.0 / MCP v2.27.0 parked on legacy-v1 for six months. Four partner integrations landed clean during the stability window. Release notes

Apr 19tractionInterop Week 1 opened — Five-signal compose test proposal. Two fixtures confirmed within 45 minutes of open (Nobulex step 2, continuity-analyzer step 3). vocab#36

Apr 19tractioncontinuity-analyzer crosswalk merged — nutstrut's PR #33 after a five-check protocol review.

Apr 18shipAgentNexus Track A round-trip — Fixtures verify end-to-end under APS. JCS + Ed25519 + delegation walk + monotonic narrowing. PR #17

Apr 18shipVeritasActa KU receipt signer — APS receipt slots into external_receipts.aps. Cross-verified. PR #7

Apr 18paperPaper 7: Cognitive Attestation — A cryptographic commitment attached to an agent's action record, to which sparse-autoencoder features were active during the decision. Envelope spec, three-stage verification, four dispute primitives. Experimental results against Llama-3.1-8B via Neuronpedia. Zenodo.

Apr 18tractionVocab registry at 14 external crosswalks — SINT refresh (PR #30) and RNWY A2A Agent Card mapping (PR #32) both merged same day.

Apr 18tractionMnemoPay composition hook — Three-point read on x402#1904. Composition via delegation-reference in X-Agent-Identity.

Apr 18tractionArkForge three-plane decomposition — +1 on desiorac's delegation/decision/execution model. Notes-column cross-reference pinned in ECOSYSTEM table. ATF#8

Apr 17tractionaeoess-aps crosswalk published — First time we published our own crosswalk alongside the twelve external partner crosswalks in the registry.

Apr 17tractionAPS added to governance_attestation.issuers_in_production — 4th production issuer via Build D2 JWS trust profiles.

Apr 16shipBuild C — Settlement Pipeline — Per-period signed settlement records aggregating Attribution Primitives across D/P/G/C axes. 3 new MCP tools, 123 modules. Blog

Apr 16shipBuild A — unified four-axis attribution primitive — One signed Merkle receipt across D (data), P (protocol), G (governance), C (compute). Four independently-verifiable projections, one envelope. SDK v2.0.0, MCP v3.1.1, Python v0.13.0. Blog.

Apr 15shipSDK v2.0.0 — Solana wallet_ref — Added to the chain enum with base58 validation. End-to-end wallet binding now spans Ethereum, Bitcoin, Solana.

Apr 15tractionVocabulary registry — four PRs merged — asqav (ML-DSA-65), JEP, insumerapi license fix, validator cleanup. 14+ contributors since Apr 11. Repo.

Apr 14paperGovernance in the Medium — paper published — Unit of agent governance is the population-with-medium, not the agent. Six rounds of adversarial review. Zenodo.

Apr 14shipThree boundary primitives — AttributionConsent, ProvisionalStatement, HumanEscalationFlag. Representation, commitment, escalation. SDK v2.0.0. Blog.

Apr 14tractionWG scope ratified — Authority Constraints + Vocabulary — 72-hour window closed on qntm#7. Six co-authors accepted with shipped-code evidence. qntm#7.

Apr 10shipPDR v2.19 adapter batch — six primitives — Six new exports close the gap between paper citations and shipped code. SDK v2.0.0. Blog.

Apr 10paperPDR in Production v2.19 cites Agent Passport System — Three-axis framework; Agent Passport System is the within-session fidelity axis. §8.10 substrate-swap is the test that would settle the orthogonality claim.

Apr 10shipAgent Governance Vocabulary repo — Neutral canonical-naming layer. Four teams converging in the first hour. Blog.

Apr 10tractionCross-vendor convergence — APS + MolTrust + AgentNexus on a shared governance_attestation envelope. APS ↔ SINT handshake v1.0-draft landed with 11 conformance tests.

Apr 9shipRepositioning — SDK leads with /core (25 curated functions), MCP defaults to APS_PROFILE=essential (20 tools). Full surface still available. Blog.

Apr 8shipQuantum governance on real hardware — 7 experiments on IBM Quantum. Bell 5.2pp, GHZ 7.7pp fidelity gaps. Blog.

Apr 8paperPaper 5: Physics-Enforced Delegation — Governing quantum hardware quality. Zenodo · Code.

Apr 8tractionExternal verification — tomjwxf verified 3/3 APS composition receipts via protect-mcp (first cross-engine). OWASP BBIS scored APS 10/12.

Apr 8paperPaper 4: Behavioral Derivation Rights — Telemetry scopes, BMOs, BYOM. Zenodo.

Apr 7shipNew protocol primitives — Bilateral completion receipts, scope_version_hash, measurementType discriminator, per-task-class trust profiles.

Apr 5ship6-session build day — Key rotation, auto-mint receipts, audit packets, 9-section governance export, trust bootstrap adapters. Blog.

Apr 4shipMicrosoft AGT approved + SINT v0.2 shipped — AGT PR#598 approved (fail-closed signature verification). SINT v0.2 shipped with our delegation_depth_floor. Blog.

Apr 4standardW3C behavioral attestation spec normative — Timing asymmetry became normative. Evidence-based passport grading + freshness semantics across 11 threads.

Apr 3shipBring Your Own Identity — did:key, did:web, SPIFFE SVID, OAuth interop. Python SDK v0.8.0 proves cross-language: signatures round-trip TS ↔ Python. Blog.

Apr 3tractionMicrosoft AGT PR fixes + production audit PASS — Fail-closed verification, dependency pinning, input validation pushed. Audit clean.

Apr 2shipSDK v2.0.0 — governance hardening — 34 new tests. 3-layer architecture convergence across OpenShell, OWASP, W3C. Blog.

Apr 1shipFirst Code Integration + 5 Security Fixes — PR merged (Solana Agent Kit). 12 features, 5 security gaps closed, compaction-drift probe, tool integrity. 29 threads, 99 modules, 2,085 tests. Blog

Mar 29shipAgent Wallets — Nano payment rail + wallet system. Feeless, delegation-scoped. Wallet · Blog

Mar 29shipPixel attribution live — Data source tracking, access receipts, derivation chains, settlement. Dashboard

Mar 29tractionCited in production paper — PDR in Production (University of British Columbia) validates APS Bayesian model. Zenodo

Mar 28rebrandAn open protocol for accountable AI agents — Academic redesign. Enterprise positioning. 10-question FAQ. Blog

Mar 27paperFaceted Authority Attenuation — Product lattice model. Seven dimensions. Zenodo

Mar 27standardIETF Internet-Draft submitted — draft-pidlisnyi-aps-00. Zero idnits errors.

Mar 26shipInstitutional governance layer — Charter, offices, approval, federation, reserves. Blog

Mar 25deployFirst publication deploys APS — Every article cryptographically governed.

Mar 25shipGovernance distribution — aps.txt, 360 consumer loop, 150 MCP tools. Blog

Mar 25shipInteractive protocol map — 57-module molecular layout. Blog

Mar 24standard3 WG specs ratified — QSP-1, DID Resolution, Entity Verification. Blog

Mar 23shipOATR founding member + data governance — Ledger, settlement, attribution. Blog

Mar 22shipConstitutional v2 complete — 32 modules. 2,085 tests. Blog

Mar 22shipFirst encrypted relay envelope — E2E through qntm bridge. Blog

Mar 21shipDecision semantics — Content-addressable decisions. Blog

Mar 20standardAMCS v0.1.0 — AI-native media credentialing spec. Spec

Mar 20shipData attribution layer — Contribution receipts, Merkle proofs. Blog

Mar 19ship8 modules in one session — Oracle witness, audit bridge, policy conflict, key rotation. Blog

Mar 17standardWG spec demand — Three groups asked for the same thing. Blog

Mar 16tractionYC CEO endorsed — Garry Tan repost. Microsoft merged APS code. Federal agency reviewing. Blog

Mar 15shipMingle v2.0 — Semantic matching, persistent identity, ghost mode. More

Mar 14shipSubstack launch — Cross-protocol bridge article. Blog

Mar 12shipMingle v1 ships — Your AI finds people for you. Blog

Mar 11shipIntent Network — Publish-discover-match for agents. Blog

Mar 10shipReputation-gated authority — Agents earn trust, not just receive it. Blog

Mar 9paperPaper 2: Monotonic Narrowing — Authority attenuation formalized. Zenodo

Mar 7shipAutoresearch — AI finds bugs AI wrote. Blog

Mar 6shipPrincipal identity + Python SDK — Three new protocol extensions. Blog

Mar 5standardOWASP AI Security mapping — Community health baseline. Blog

Mar 4shipSDK v1.21.2 + MCP v2.12.0 — Two agents get their next mission. Blog

Mar 3shipFirst real audit — Agents review the code. Blog

Mar 2shipGraduated enforcement + threat model — Agent District RPG. Blog

Mar 1shipAgentic commerce — Layer 8. 4-gate checkout. MCP v2.1.0. Blog

Feb 28shipDocumentation sprint — llms.txt, passport spec. Blog

Feb 27shipCoordination primitives — Task lifecycle, evidence, review. Blog

Feb 25shipIntent architecture — Layer 5 foundations. Blog

Feb 21paperPaper 1: The Agent Social Contract — First formalization. Zenodo

Feb 18startProject begins — Ed25519 identity, delegation chains, first tests.

What is Agent Passport System

What does the protocol do?

Agent Passport System 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.

How is it different from other agent frameworks?

Most frameworks handle orchestration. Agent Passport System handles enforcement: what can this agent do, and what happens when it violates a constraint? Bring your own identity (did:key, did:web, SPIFFE, OAuth, did:aps). The gateway is both judge and executor. Monotonic narrowing, cascade revocation, earned reputation, values floor enforcement, 4-gate commerce, agent-to-agent payments, Merkle-proven attribution, cross-DID-method composition, mutual authentication, instruction-provenance receipts. Policy eval under 2ms. 2,884 tests. Works with any framework.

Is this production-ready?

Policy evaluation in under 2ms. 403 ops/sec sustained throughput. Sub-millisecond denial. 14 constraint dimensions checked per action. 2,884 tests. Eight published papers. An IETF Internet-Draft (draft-pidlisnyi-aps-00). 20 essential MCP tools (150 total). SDK on npm and PyPI. tens of thousands of installs. Independently cited by PDR in Production (University of British Columbia). Apache-2.0 licensed.

Who is this for?

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.

How does delegation work?

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.

How do I integrate with my existing framework?

Agent Passport System works alongside any agent framework — CrewAI, LangChain, A2A, ADK, or custom. The SDK provides the identity and governance layer; your framework handles orchestration. The MCP server exposes every protocol operation — 20 essential tools by default, full surface under APS_PROFILE=full. Install the SDK, issue a passport, attach it to your agent. The framework doesn't need to change.

How does revocation work at scale?

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.

Is there a standard?

Yes. draft-pidlisnyi-aps-00 is submitted to the IETF. Eight 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.

What about compliance?

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.

What's the pricing?

The protocol and SDK are free and open source (Apache-2.0). Always will be. Agent-to-agent payments use the agent wallet (currently Nano-based, multi-chain in roadmap): sub-second, delegation-scoped. The hosted enforcement gateway is live: free plan (3 agents, 1K evaluations/mo), Team $99/mo (25 agents, 50K evaluations), Enterprise custom. Sign up at portal.

For AI agents: Visit AGENTS.md for agent-friendly project instructions, llms.txt for machine-readable documentation, llms-full.txt for the full technical reference, or .well-known/mcp.json for MCP server discovery. This page is designed for humans.