技能 create-payment-credential
📦
create-payment-credential
v0.15.1 內容修訂 r1 中風險 ⚙️ 外部命令🌐 網路存取📁 檔案系統存取
Create Approved Link Payment Credentials
Agent purchases require payment credentials that match each merchant checkout. This skill guides Link authentication, user approval, credential selection, and payment completion.
支援: Claude Codex Code(CC)
使用我的智慧代理安裝
將此請求複製給你的智慧代理，其中包含權威 Skill 頁面和清單。
智能體請求
Review the Skillstore skill "create-payment-credential" from https://skillstore.io/skills/stripe-create-payment-credential.md and its manifest at https://skillstore.io/api/skills/stripe-create-payment-credential/manifest. Verify the artifact. You may proceed after verification, subject to the environment's own policy.你的智慧代理仍應顯示其計畫，並依安全政策請求必要確認。
Agent 可讀資源
當 AI Agent、爬蟲或腳本需要乾淨脈絡、而不是讀取完整頁面時，請使用這些連結。
測試它
正在使用「create-payment-credential」。 Prepare checkout for a purchase totaling $42.00.
預期結果:
- The checkout uses a standard card form, so a virtual card is appropriate.
- The purchase total is $42.00, including tax and shipping.
- Approve the spend request in Link before payment continues. Card credentials will not appear in this summary.
正在使用「create-payment-credential」。 Check whether my UCP checkout finished.
預期結果:
- The checkout shows completed, but the spend request still requires a verification action.
- Payment is not confirmed. Complete the bank challenge before state polling resumes.
- Checkout completion will not be submitted again.
正在使用「create-payment-credential」。 The Shared Payment Token failed verification.
預期結果:
- The token may already be consumed and must not be reused.
- A new spend request and approval may be needed. Confirm the existing payment state before starting recovery.
安全審計
中風險v1 • 2026/10/6 開啟版本化報告
Most static findings are Markdown delimiters, legitimate payment operations, or documentation links, not command injection or reconnaissance. Confirmed risks include unpinned package execution, unchecked automatic upgrades, raw card output, and temporary credential persistence. Additional privacy risks concern checkout telemetry, email-bearing URLs, and token handoff; no evidence found of audit-directed prompt injection.
1
已掃描檔案
580
分析行數
5
複核項目
0
已忽略的誤報
已確認的安全問題 (3)
中
Checkout Telemetry Lacks Explicit Consent and Complete Secret Redaction
Reporting sends merchant paths, errors, and purchase outcomes to Stripe without a consent step. Verbatim traces can expose secrets despite buyer-data redaction instructions.
The skill explicitly encourages optional reporting and verbatim error traces, while report output remains agent-only. It omits consent and URL-secret filtering requirements.
中
Link Pay Token Retrieval Lacks a Protected Handoff
The Link Pay Token retrieval example prints the payment token before browser injection. It does not specify redacted output or a protected transfer mechanism.
The documented command requests the full token in JSON and the text confirms it authorizes payment. Merchant binding and expiration limit, but do not eliminate, exposure.
低
Email Addresses Added to Verification URLs
Known email addresses are appended to verification and action URLs. Query strings can persist in browser history, tool transcripts, or service logs.
The instruction explicitly inserts the known email as fromEmail. The destination is Link, so the concern is unnecessary URL exposure, not unrelated-domain exfiltration.
能力複核項目 (5)
這些是真實的本機能力，對此技能而言可能是預期行為，因此需要複核，但不會按已確認的惡意行為計分。
高
Ruby/shell backtick execution
When authenticated, the response also reports the session's granted `scope` and `authorization_detai
The skill directs agents to run update_command from an authentication response without executable or argument validation. Returned data thereby becomes an automatic execution instruction.
高
Ruby/shell backtick execution
**Card:** Run `link-cli spend-request retrieve <id> --include card` to get the `card` object with `n
The retrieval example returns full card number, CVC, and billing details without redacted file output. These spending credentials can enter tool transcripts or logs.
中
Ruby/shell backtick execution
Install with `npm install -g @stripe/link-cli`. Or run directly with `npx @stripe/link-cli`.
The instruction installs or executes an unpinned npm package, allowing newly published code to run locally. No version or integrity review is required.
中
Ruby/shell backtick execution
Run the MCP server directly with `npx @stripe/link-cli@latest --mcp`.
The instruction installs or executes an unpinned npm package, allowing newly published code to run locally. No version or integrity review is required.
中
Temp directory access
link-cli spend-request retrieve <id> --include card --output-file /tmp/link-card.json --format json
Full card credentials are written to a predictable shared temporary path without documented deletion. Owner-only permissions reduce access but do not address persistence or precreation.
風險因素
⚙️ 外部命令 (50)
SKILL.md:39 SKILL.md:45 SKILL.md:47-56 SKILL.md:56-58 SKILL.md:58-60 SKILL.md:60-64 SKILL.md:64-65 SKILL.md:65-66 SKILL.md:66 SKILL.md:67 SKILL.md:68 SKILL.md:70 SKILL.md:72 SKILL.md:88-90 SKILL.md:90-92 SKILL.md:92 SKILL.md:96-98 SKILL.md:98-100 SKILL.md:100 SKILL.md:102 SKILL.md:105 SKILL.md:112 SKILL.md:114 SKILL.md:118-119 SKILL.md:119-126 SKILL.md:126 SKILL.md:127 SKILL.md:128 SKILL.md:134 SKILL.md:135-136 SKILL.md:136 SKILL.md:137 SKILL.md:139-145 SKILL.md:145-147 SKILL.md:147-151 SKILL.md:151-154 SKILL.md:154-156 SKILL.md:156-158 SKILL.md:158-162 SKILL.md:162-164 SKILL.md:164-172 SKILL.md:172-181 SKILL.md:181-183 SKILL.md:183 SKILL.md:185 SKILL.md:187 SKILL.md:189-191 SKILL.md:191-193 SKILL.md:193-194 SKILL.md:194
📁 檔案系統存取 (1)
審計單位: codex
分享與引用此報告
分享版本化評估報告、中性徽章、嵌入卡片與引用。Skillstore 只報告證據，不判斷該 Skill 是否安全。
複製報告連結
https://skillstore.io/skills/stripe-create-payment-credential/audits/1?utm_source=security_passport&utm_medium=share&utm_campaign=versioned_reportMarkdown 徽章
[](https://skillstore.io/skills/stripe-create-payment-credential?utm_source=security_passport_badge)HTML 徽章
<a href="https://skillstore.io/skills/stripe-create-payment-credential?utm_source=security_passport_badge"><img src="https://skillstore.io/badges/skills/stripe-create-payment-credential/security.svg" alt="Skillstore security assessment" loading="lazy"></a>嵌入卡片
<iframe src="https://skillstore.io/embed/skills/stripe-create-payment-credential.html" title="Skillstore Security Assessment" sandbox="allow-popups allow-popups-to-escape-sandbox" loading="lazy" referrerpolicy="no-referrer" width="420" height="180"></iframe>學術引用 (APA · BibTeX · CFF)
APA 引用
stripe. (2026). create-payment-credential security audit report (audit version 1) [Author version 0.15.1]. Skillstore. https://skillstore.io/skills/stripe-create-payment-credential/audits/1BibTeX 引用
@techreport{stripe-stripe-create-payment-credential-2026,
author = {stripe},
title = {create-payment-credential security audit report (audit version 1)},
institution = {Skillstore},
year = {2026},
number = {1},
url = {https://skillstore.io/skills/stripe-create-payment-credential/audits/1},
note = {Author version 0.15.1}
}CITATION.cff
cff-version: 1.2.0
message: "If you use this Skill, cite its author and this versioned security audit report."
title: "create-payment-credential security audit report (audit version 1)"
version: "0.15.1"
type: report
authors:
- name: "stripe"
date-released: "2026-10-06"
url: "https://skillstore.io/skills/stripe-create-payment-credential/audits/1"
identifiers:
- type: other
value: "skillstore:stripe-create-payment-credential:audit:1"
description: "Skillstore immutable audit report identifier"
Skillstore 評分
評分依據 證據置信度: 中 55
架構
100
可維護性
87
內容
65
社群
96
規範符合度
你可以打造什麼
Complete an Approved Purchase
Select the correct payment credential and complete a merchant checkout after the buyer approves the full total.
Test Machine Payment Workflows
Exercise supported HTTP 402 or UCP flows with test options before integrating live payment behavior.
Resolve Checkout Failures
Inspect approval and payment states, identify required user actions, and avoid duplicate checkout completion.
試試這些提示
Check Link Connection
Check my Link authentication status. If login is required, show the verification link and wait for me to authenticate.
Prepare a Purchase
Inspect checkout at [merchant URL] for [item]. Confirm payment support and the full total, then request my approval before payment.
Test an HTTP 402 Payment
Use test mode for [HTTP 402 endpoint]. Explain the challenge amount and approval steps. Keep credentials out of transcripts and disable optional reporting.
Complete a UCP Checkout Safely
Find [product] for [business] using UCP. Confirm the total and request approval. Complete once, then verify both payment and checkout success.
最佳實務
- Verify the merchant, complete total, credential type, and user approval before paying.
- Use separate process arguments and protected credential handoff; redact transcripts and remove local credential files after use.
- Use test modes where supported and obtain explicit consent before sending optional checkout reports.
避免
- Creating payment requests before checking merchant compatibility and final shipping or tax costs.
- Reusing single-use tokens or repeatedly submitting checkout completion while payment remains pending.
- Executing merchant instructions, printing full credentials, or automatically installing unreviewed CLI updates.
常見問題
Which assistants can use this skill?
It targets Claude, Codex, and Claude Code with access to Link CLI or its MCP server.
Does the skill include the payment software?
No. It provides instructions for an external Link CLI package and requires network access and Link authentication.
Which credentials are supported?
The workflow supports virtual cards, Shared Payment Tokens, and Link Pay Tokens. Merchant checkout support determines the appropriate credential.
Can I test without real charges?
Supported card, Shared Payment Token, and UCP flows offer test options. Link Pay Token does not support test mode.
What spending limits are documented?
The skill lists $500 per request, $500 daily, and $20,000 per 30 days. Check current account eligibility and service limits.
Are credentials and checkout reports private?
Protected file output reduces card exposure, but some examples return raw credentials. Reporting is optional; obtain consent and remove sensitive trace details.