ar-agents is a Mercado Pago Agent Toolkit for the Vercel AI SDK 6. It ships 89 typed tools that an LLM agent can call directly to drive Mercado Pago billing flows: subscriptions, payments, refunds, checkout pro, marketplace OAuth, cuotas (installments), QR in-store, 3DS challenge resolution, point-of-sale devices, webhooks. Sidecar packages cover AFIP/ARCA, WhatsApp Business Cloud, banking (CBU/CVU + BCRA), and shipping (Andreani/OCA/Correo Argentino).

How is this different from the official mercadopago SDK?

The official mercadopago SDK is a thin REST client. It does not ship Vercel AI SDK tool schemas, does not implement webhook HMAC verification with replay protection, does not run on Edge Runtime (Node-only), and does not gate irreversible operations. ar-agents adds all of that on top of the underlying API: 89 typed tools, deterministic idempotency keys derived from inputs, programmatic human-in-the-loop on refund/cancel/delete, npm provenance attestation, Vercel KV adapters via subpath, OpenTelemetry instrumentation. You can use both packages in the same project; ar-agents wraps the underlying API directly without depending on the official SDK.

Does ar-agents work on Edge Runtime?

Yes. The whole package is Web Crypto-based with no node:crypto dependency, so it runs on Vercel Edge Functions, Cloudflare Workers, Deno, and any other V8-isolate runtime. Webhook signature verification, HMAC, and idempotency-key generation all use the Web Crypto API.

What is HITL (human-in-the-loop) in ar-agents?

Eight tools mutate state irreversibly (refund_payment, cancel_subscription, delete_customer_card, etc.). The toolkit accepts a requireConfirmation callback that gates each invocation: the tool function literally will not execute until your callback returns true. This is a programmatic gate, not just an LLM instruction. You can show the user a UI and wait for their explicit approval before any irreversible operation runs.

How does idempotency work?

Every POST request gets an auto-generated idempotency key. For LLM-driven retries, four mutating tools (create_payment, create_subscription, create_payment_preference, refund_payment) use a deterministic key derived from a SHA-256 hash of the meaningful inputs (external_reference, amount, payment_method, etc.). Same inputs produce the same key, so retries return the existing resource instead of double-charging the customer.

Yes. MIT license. No paid tier, no telemetry phone-home, no usage caps. The package is published to npm under the @ar-agents scope with SLSA v1 provenance attestations.

What about AFIP, WhatsApp, banking, shipping?

Sidecar packages cover the rest of the Argentine business stack: @ar-agents/identity (CUIT/CUIL validation + AFIP/ARCA padron lookup with monotributo category and IVA condition), @ar-agents/facturacion (AFIP/ARCA factura electronica via WSFE), @ar-agents/whatsapp (WhatsApp Business Cloud API with HMAC webhook verify and AR phone normalizer), @ar-agents/banking (CBU/CVU validation + BCRA Central de Deudores), @ar-agents/shipping (Andreani, OCA, Correo Argentino), @ar-agents/identity-attest (HMAC-signed verification orchestrator). Each ships independently to npm.

Is there a Model Context Protocol (MCP) server?

Yes. @ar-agents/mcp bundles all 7 ar-agents packages into a single MCP server compatible with Claude Desktop, Cursor, Codeium, Continue, Cline, or any MCP host. Auto-detects which packages to enable from environment variables. Listed on Glama (glama.ai/mcp/servers/ar-agents/ar-agents) and the official MCP Registry (io.github.ar-agents/mcp).

/refs · BibTeX · APA · Chicago

Academic references.

Pre-formatted citation entries for every published artifact: the five RFCs (each deposited on Zenodo with its own DOI), the test-vectors dataset, the open-source toolkit. For comparative-law scholars + AI-governance researchers + anyone citing this work in literature. All under CC-BY-4.0 (specs) or MIT (code), re-use freely with attribution.

RFC-001: Three-layer civil liability framework for Argentine AI corporations

rfc-001-v1 · rfc

BibTeX

@techreport{clemente2026rfc001,
  author      = {Naza},
  title       = {{RFC-001}: Three-layer civil liability framework for Argentine AI corporations},
  institution = {ar-agents},
  year        = {2026},
  type        = {{Request for Comments}},
  number      = {001},
  version     = {v1},
  doi         = {10.5281/zenodo.20159396},
  url         = {https://doi.org/10.5281/zenodo.20159396},
  note        = {CC-BY-4.0. Canonical: https://ar-agents.ar/rfcs/001}
}

APA

Naza (2026). RFC-001: Three-layer civil liability framework for Argentine AI corporations (Request for Comments No. 001, Version 1). ar-agents. https://doi.org/10.5281/zenodo.20159396

Chicago

Naza. "RFC-001: Three-layer civil liability framework for Argentine AI corporations." Request for Comments 001, ar-agents, 2026. https://doi.org/10.5281/zenodo.20159396.

RFC-002: Agent-discovery-by-default convention

rfc-002-v1 · rfc

BibTeX

@techreport{clemente2026rfc002,
  author      = {Naza},
  title       = {{RFC-002}: Agent-discovery-by-default convention},
  institution = {ar-agents},
  year        = {2026},
  type        = {{Request for Comments}},
  number      = {002},
  version     = {v1},
  doi         = {10.5281/zenodo.20159407},
  url         = {https://doi.org/10.5281/zenodo.20159407},
  note        = {CC-BY-4.0. Canonical: https://ar-agents.ar/rfcs/002}
}

APA

Naza (2026). RFC-002: Agent-discovery-by-default convention (Request for Comments No. 002, Version 1). ar-agents. https://doi.org/10.5281/zenodo.20159407

Chicago

Naza. "RFC-002: Agent-discovery-by-default convention." Request for Comments 002, ar-agents, 2026. https://doi.org/10.5281/zenodo.20159407.

RFC-003: Cross-jurisdictional audit-log reciprocity

rfc-003-draft · rfc

BibTeX

@techreport{clemente2026rfc003,
  author      = {Naza},
  title       = {{RFC-003}: Cross-jurisdictional audit-log reciprocity},
  institution = {ar-agents},
  year        = {2026},
  type        = {{Request for Comments}},
  number      = {003},
  version     = {draft},
  doi         = {10.5281/zenodo.20159411},
  url         = {https://doi.org/10.5281/zenodo.20159411},
  note        = {CC-BY-4.0, draft. Canonical: https://ar-agents.ar/rfcs/003}
}

APA

Naza (2026). RFC-003: Cross-jurisdictional audit-log reciprocity (Request for Comments No. 003, Draft). ar-agents. https://doi.org/10.5281/zenodo.20159411

Chicago

Naza. "RFC-003: Cross-jurisdictional audit-log reciprocity." Request for Comments 003 (draft), ar-agents, 2026. https://doi.org/10.5281/zenodo.20159411.

RFC-004: Operational-log specification for AR AI corporations

rfc-004-draft · rfc

BibTeX

@techreport{clemente2026rfc004,
  author      = {Naza},
  title       = {{RFC-004}: Operational-log specification for AR AI corporations},
  institution = {ar-agents},
  year        = {2026},
  type        = {{Request for Comments}},
  number      = {004},
  version     = {draft},
  doi         = {10.5281/zenodo.20159417},
  url         = {https://doi.org/10.5281/zenodo.20159417},
  note        = {CC-BY-4.0, draft. Canonical: https://ar-agents.ar/rfcs/004}
}

APA

Naza (2026). RFC-004: Operational-log specification for AR AI corporations (Request for Comments No. 004, Draft). ar-agents. https://doi.org/10.5281/zenodo.20159417

Chicago

Naza. "RFC-004: Operational-log specification for AR AI corporations." Request for Comments 004 (draft), ar-agents, 2026. https://doi.org/10.5281/zenodo.20159417.

RFC-005: Asymmetric upgrade for the operational log

rfc-005-draft · rfc

BibTeX

@techreport{clemente2026rfc005,
  author      = {Naza},
  title       = {{RFC-005}: Asymmetric upgrade for the operational log},
  institution = {ar-agents},
  year        = {2026},
  type        = {{Request for Comments}},
  number      = {005},
  version     = {draft},
  doi         = {10.5281/zenodo.20159424},
  url         = {https://doi.org/10.5281/zenodo.20159424},
  note        = {CC-BY-4.0, draft. Canonical: https://ar-agents.ar/rfcs/005}
}

APA

Naza (2026). RFC-005: Asymmetric upgrade for the operational log (Request for Comments No. 005, Draft). ar-agents. https://doi.org/10.5281/zenodo.20159424

Chicago

Naza. "RFC-005: Asymmetric upgrade for the operational log." Request for Comments 005 (draft), ar-agents, 2026. https://doi.org/10.5281/zenodo.20159424.

RFC-004 v1 conformance test vectors

test-vectors-rfc-004-v1 · dataset

BibTeX

@misc{clemente2026testvectors,
  author       = {Naza},
  title        = {{RFC-004 v1} conformance test vectors},
  year         = {2026},
  publisher    = {ar-agents},
  howpublished = {\url{https://ar-agents.ar/test-vectors/rfc-004-v1.json}},
  note         = {Dataset, CC-BY-4.0}
}

APA

Naza (2026). RFC-004 v1 conformance test vectors [Dataset]. ar-agents. https://ar-agents.ar/test-vectors/rfc-004-v1.json

Chicago

Naza. "RFC-004 v1 conformance test vectors." Dataset, ar-agents, 2026. https://ar-agents.ar/test-vectors/rfc-004-v1.json.

ar-agents: Open-source infrastructure for Argentine AI corporations

ar-agents-toolkit · software

BibTeX

@software{clemente2026aragents,
  author       = {Naza},
  title        = {{ar-agents}: Open-source infrastructure for {Argentine} {AI} corporations},
  year         = {2026},
  publisher    = {GitHub},
  url          = {https://github.com/ar-agents/ar-agents},
  note         = {MIT license}
}

APA

Naza (2026). ar-agents: Open-source infrastructure for Argentine AI corporations [Computer software]. GitHub. https://github.com/ar-agents/ar-agents

Chicago

Naza. ar-agents: Open-source infrastructure for Argentine AI corporations. v0.x, 2026. https://github.com/ar-agents/ar-agents.

Notes on citation

Cite the RFC by version (v1) if you reference the stable release; by "draft" if the document is still in active revision (currently RFC-003 + RFC-004). The frozen v1 vectors at /test-vectors stay stable so legislation referencing them has a consistent target.
For replication studies of the conformance vectors, cite the specific JSON file URL + the publishedAt date inside the file.
For software citation, prefer CITATION.cff (planned next at github.com/ar-agents/ar-agents/CITATION.cff).