Head to head · Product search for shoppers · October 2026 research run

opencommerce vs Zinc

Zinc scores 69.9 (B) on agent readiness against opencommerce's 57.9 (C), and leads in 5 of 7 scored categories. opencommerce leads on security & auth. Both do product search for shoppers.

Best shopping and order APIs for AI agents buying for a person · All 50 shopping comparisons

Which one, for what

opencommerce C

Best for A company building a chat agent that buys for US shoppers at many online stores, restaurants and airlines through one tool, with payments approved by the shopper's passkey.

Ahead on

  • Security & auth, 68 against 57

Watch for

The tool description is 57 rules and about 14,500 characters, read on every call

Zinc B

Best for Agents that need to buy physical goods from US retailers for a person, with machine payment and no account.

Ahead on

  • Reliability, 55 against 25
  • Schema & documentation, 92 against 81
  • Agent ergonomics, 86 against 69
  • Payments & pricing, 98 against 60
  • Maintenance & community, 83 against 75

Also in its favour

  • Free to start without a card
  • No incidents deducted, where opencommerce loses 3 points for them

Watch for

By default orders run on retailer accounts Zinc runs, and its docs index lists a guide to Amazon's anti-bot pages and one on proxies "for account protection"

Why the edge

Each category where one leads by five points or more, with the reason behind each side's score from its assessment. The numbers are in the table below; this is what they stand for.

Payments & pricing Zinc +38

opencommerce 60

No x402, MPP or L402. Purchases are paid with the shopper's card from the Vault, and agentcard.com's FAQ answers Not yet on agent payment protocols (0 of 40). 1 per cent of each completed order is published on opencommerce.sh without a login (20 of 20). Sandbox credentials run the whole flow, stop before payment and are never billed, with no card needed (20 of 20). A person signs in through app.agentcard.sh or an emailed code, and since 27 October 2026 the CLI says sign-ups are by invitation (0 of 20). Total 40. Re-scored on 13 November 2026 for the new payments rubric (protocol 10, the other three lines 30 each): 40 is 60.

Zinc 98

x402 (v2, USDC on Base and Solana) and MPP (Tempo, and Link with a Shared Payment Token) on POST /agent/orders and the /agent/* data endpoints. Returns, cases, managed accounts and webhooks need an API key (8 of 10). $1 per successful order plus the item, $0.25 for tracking links and emails, $0.10 per item over 10, and $0.01 per data call, published without login (30). A sandbox needs no account or card, and a device-approved live key comes with a $5 starter credit (30). An agent can pay per order with no account, or mint a sandbox key with one unauthenticated call (30). 8 + 30 + 30 + 30 = 98.

Reliability Zinc +30

opencommerce 25

Read on the local-package lines for its listed kind, sdk. The official npm package @agent-cards/commerce-sdk runs on Node 18 or later and the browser, as the quickstart states (20). The repository the package names is private and no public CI or test suite was found. The only public repository is the example app, tiny-agent-company/commerce-sdk-example, with four commits on 29 October 2026 and no workflows (0). There is no public issue tracker, and reports go to agent-cards support chat and support@agentcard.sh, so the crash and regression line has nothing to read (0). Versions are 0.x semver on npm, deprecated names such as toAnthropicTool are kept with notes, and Agentcard's shared changelog marks breaking platform changes, but the SDK has no changelog of its own (5 of 15). 0.18.6 per the 30 October check, not declared stable (0). 20 + 0 + 0 + 5 + 0 = 25.

Zinc 55

Zinc's own status page (www.zinc.com/status) shows five components (Ordering, Tracking, Product Data, API, Dashboard) with 90-day uptime bars (20). On 18 November 2026 it showed 99.1 per cent uptime for Ordering and 99.57 per cent for the API over 90 days, both recovered from a degradation that day. The page has no incident write-ups, and 99.1 per cent over 90 days is about 19 hours, so at least one major interruption is implied (10 of 30). No numeric request rate limits found, apart from 5 sandbox key mints per IP address a day (5 of 15). 429 carries rate_limit_exceeded with guidance to back off and retry, responses carry RateLimit and RateLimit-Policy headers (changelog, August 2026), and orders and top-ups take idempotency keys (15). No SLA found (0). Orders are generally available. Cross-retailer search and the Zinc 2.0 announcement carry a beta label (5 of 10). 20 + 10 + 5 + 15 + 0 + 5 = 55.

Agent ergonomics Zinc +17

opencommerce 69

One tool, inside the band for ten or fewer, but its description is 57 rules and about 14,500 characters, and the docs also recommend Agentcard's appInstructions() system prompt. Answers are capped at eight product and eight store cards and stripped of links for the model. Ten less than the band for the description's size, two back for the capped answers, a departure from the checklist (17 of 25). The buy tool has no page or filter controls, and storeWaitMs sets how long a reply waits. GET /buy/merchants has q, kind, limit up to 100 and a cursor (12 of 20). AgentcardApiError codes each come with what to do next, decline_code and error_code are machine-readable, and input outside the schema is refused with a message the model can act on (18 of 20). No idempotency key on a purchase turn. The client never resends a turn except once after a 401, a cart hash guards confirms, order events carry an order_id to reconcile on, and the MCP buy tool is documented as destructive (14 of 20). The model fills five simple fields, but the SDK is TypeScript only, for Node 18 or later and the browser (8 of 15). Total 69.

Zinc 86

Search returns up to 20 results, best first, and data and order calls are separate routes, so responses stay small. No field selection found (20 of 25). List and export endpoints take filters, and orders can be listed, exported and followed by webhook (18 of 20). Errors carry a stable code, a message and details, including details.fund_with on insufficient_funds and RFC 8628 hints in the device flow (20). Idempotency keys on orders, agent orders and top-ups, a 409 that names the existing order, and an MCP spending tool that needs a confirmation token only a person's click produces. MCP annotations weren't read (18 of 20). Few required fields, and an official npm SDK (zinc, CalVer from the live spec). No second official SDK was confirmed (10 of 15). 20 + 18 + 20 + 18 + 10 = 86.

Schema & documentation Zinc +11

opencommerce 81

The buy tool's input is a strict JSON Schema with five required fields and additionalProperties false, and the package ships TypeScript declarations. The purchase endpoints under it are missing from the OpenAPI file at docs.agentcard.com/openapi.json, which holds 18 operations for tokens, connections, identity checks and cards (20 of 25). llms.txt on opencommerce.sh and on docs.agentcard.com, with a Markdown twin for each docs page (10 of 10). The description's 57 rules say when to call the tool, when to set confirm and what each answer means, such as "Show the shopper the cart before you confirm it" (18 of 20). Two booleans and two nullable ids, but the main input ask is free text (9 of 15). Code for Anthropic, Vercel AI SDK, OpenAI and OpenRouter, a sandbox walkthrough script, error tables with what to do next and captured error bodies (14 of 15). npm versions follow 0.x semver, and Agentcard's changelog is dated almost daily, newest 27 October 2026, but it covers every Agentcard product and the SDK has no changelog of its own (10 of 15). Total 81.

Zinc 92

A public OpenAPI 3.1 file at api.zinc.com/openapi.json, version 2026-10-30, with 45 paths and 53 operations, 49 of them described (25). llms.txt, llms-full.txt, section indexes and Accept: text/markdown on every page (10). Pages say when to use each route, such as checking a retailer before ordering and using the offers endpoint rather than product details for a price (17 of 20). Typed request bodies with required fields, prices in integer cents and enumerated payment methods (14 of 15). Examples on every page and a long table of error codes with recovery steps, but the metered data endpoints return a different 402 shape and the MCP connector's tool schemas weren't read (11 of 15). Dated API versions (CalVer, 30 versions listed back to 24 November 2025) and a weekly changelog (15). 25 + 10 + 17 + 14 + 11 + 15 = 92.

Security & auth opencommerce +11

opencommerce 68

OAuth client credentials exchanged for a one-hour platform token, separate sandbox and production clients, and a per-shopper access token as an alternative. Agentcard's platform tokens carry the single scope api, per the agentcard listing, and init writes production credentials to the env file by default (22 of 30). Every payment not covered by an auto-approval rule needs the shopper's passkey, the docs say neither the app nor the model approves for the shopper, and the model cannot change the shopper or address. init runs Agentcard's coding agent with permissions bypassed and unsandboxed commands, which the docs disclose. Two less for that (15 of 20). Answers carry store and product text, links are stripped before the model reads them, and preset rules ignore merchant text an agent sends, but no guidance naming untrusted content or prompt injection was found (10 of 15). Order and purchase events by webhook or stream, with logged deliveries and stored approval records of who, when and where (11 of 15). agentcard.com serves a valid security.txt, the Vault Enterprise plan lists a SOC 2 report and pen-test summary, and the Trust Centre was not read. No bug bounty or disclosure policy found (10 of 20). Total 68.

Zinc 57

Bearer API keys, zn_test_ for sandbox and zn_live_ for live, and each agent can get its own live key by device-code approval. No scopes or rotation steps were found (18 of 30). Sandbox keys can't spend, every order carries a max_price cap, a person approves a live key in the browser, and the MCP connector's place-order needs a confirmation token withheld from the model (17 of 20). The MCP docs explain that the token is there so a prompt injection in a page can't become a purchase. Product data from retailers is untrusted text (12 of 15). Order timelines, a usage endpoint, CSV export and the dashboard show every order (10 of 15). No security.txt (404), disclosure policy, bug bounty, SOC 2 or ISO 27001 found (0). 18 + 17 + 12 + 10 + 0 = 57.

Maintenance & community Zinc +8

opencommerce 75

@agent-cards/commerce-sdk 0.18.6 was published on 30 October 2026, per the 30 October check, not re-read on 2 November (30 of 30). 19 versions from 16 to 30 October 2026 (20 of 20). Closed service. Agentcard's changelog is dated almost daily, newest 27 October on 2 November too, and support runs through agent-cards support chat and support@agentcard.sh. Neither was tested (10 of 15). The TypeScript SDK is current, with no SDK in a second language (10 of 15). The SDK has no dependencies. The CLI has 10, among them PostHog and the Claude Agent SDK, needs Node 22, and its repository is private, so CI isn't visible (5 of 10). The product is two weeks old. Total 75.

Zinc 83

The changelog's latest entry is dated 13 November 2026, and the npm SDK's latest version is 2026.10.30 (30). Weekly dated changelog entries, 13 since 20 August 2026 (20). A public weekly changelog and a support address (support@zinc.com) that the contact page says gets a person. GitHub issue response wasn't checked (10 of 25). The MCP connector is in the official MCP registry as com.zinc.www/mcp, published 28 October 2026 under Zinc's domain namespace, and the npm SDK is current (15). The npm SDK is generated from the live spec and published with provenance attestations (8 of 10). 30 + 20 + 10 + 15 + 8 = 83.

Score by category

CategoryWeightopencommerceZincEdge
Reliability16%2555Zinc +30
Performance10%pendingpendingpendingnot scored in this run
Schema & documentation13%8192Zinc +11
Agent ergonomics13%6986Zinc +17
Security & auth14%6857opencommerce +11
Payments & pricing10%6098Zinc +38
Task success10%pendingpendingpendingnot scored in this run
Maintenance & community7%7583Zinc +8
Transparency & trust7%6360opencommerce +3
Negative events≤15-30
TotalC 57.9/100B 69.9/100

Facts side by side

FactopencommerceZinc
KindSDK + MCPHTTP API
VendorAgentcardZinc
Hosted endpointhttps://api.agentcard.shhttps://mcp.zinc.com/mcp
TransportsHTTPStreamable HTTP
AuthOAuthOAuth or key
PricingPay per usePay per use
x402noon some endpoints
LicenceProprietary. @agent-cards/commerce-sdk and the agent-cards CLI ship a licence of Agentcard Corporation that allows running them only to use Agentcard's services and forbids copying, reverse engineering and benchmarking a competing product. The repository they name is privateProprietary service, agent skills MIT
Tools exposed18
Read-only variant documentednono
llms.txtyesyes
MCP registrynot listedcom.zinc.www/mcp
Last release2026-10-302026-11-13
Terms last updated2026-07-102026-10-12
Privacy policy last updated2026-10-212026-10-06
Customer content may train modelsnot found in the textnot found in the text
Terms restrict automated accessyesnot found in the text
Terms restrict benchmarkingnot found in the textyes
Terms or service can change without noticenot found in the textnot found in the text
Arbitration or class-action waiveryesyes
Popularity2.7k npm/wknone

How the scores have moved

Verdicts

opencommerce

One strictly typed buy tool covers search, cart, checkout and order updates, and every payment not covered by an auto-approval rule waits for the shopper's passkey. The SDK is closed source and 0.x with no public CI or issue tracker, its tool description runs to about 14.5 KB, and the purchase API under it is undocumented.

Zinc

An agent can mint a sandbox key with no signup and pay for real orders by x402, MPP or Link with no account, at $1 an order plus the item, with idempotency keys and a public OpenAPI file. Orders run on Zinc's own retailer accounts by default, and no security programme or numeric rate limits were found.

Before you call either

opencommerce

  1. Call buy with the shopper's words in ask, and set confirm true only after the shopper says yes to the cart you showed
  2. After a timeout, unreachable or invalid_answer error, read the conversation before sending the turn again, since the order may have gone through
  3. Show the cart's max_charge beside its total, because the approval holds up to that amount
  4. Use sandbox credentials for tests. init writes production credentials by default, and they place real orders
  5. Check a store with GET /buy/merchants before sending a shopper there, and suggest another store when it is held_back or supported is false

Zinc

  1. Call GET /retailers/check with the product URL and country before ordering, since unsupported retailers and countries fail predictably
  2. Set max_price and an idempotency_key on every order. If a retry returns 409, read the existing order id from error.details.order_id
  3. If your HTTP client merges WWW-Authenticate headers, add ?method=x402, ?method=tempo or ?method=stripe to get one payment challenge
  4. Store the X-Api-Key returned with a paid agent order. It is the only way to follow that order without paying again
  5. Persist a sandbox key across runs. It expires after 7 days without use, and minting is limited to 5 a day per IP address

Questions

Which is better for AI agents, opencommerce or Zinc?

Zinc scores 69.9 (B) on agent readiness against opencommerce's 57.9 (C), and leads in 5 of 7 scored categories. opencommerce leads on security & auth.

Can an agent call opencommerce and Zinc without installing anything?

Yes. opencommerce has a hosted endpoint at https://api.agentcard.sh and Zinc at https://mcp.zinc.com/mcp.

Other comparisons with opencommerce or Zinc

Machine-readable

For companies

Do agents find, use and choose your tools?

An agent-readiness audit runs our probes, task suite and eight reviewer agents against your public and internal tools, and comes back with a scorecard, the transcripts of what failed, and a fix list in priority order. From $2,500, re-run included. We never take payment to move a rank. We do help companies earn one.