CategoryWeightCategory scoreContribution

Securityprompt, exec, net, exfil, eval

35%

30.8 pts

Supply chainhash, typosquat, maintainer, lockfile

20%

100

20.0 pts

Maintenancestaleness, pinning, CI

15%

100

15.0 pts

TransparencySKILL.md, perms, README

15%

100

15.0 pts

Communityinstalls, verify, response

15%

100

15.0 pts

Findings & checks · 1 flagged

Securityscore 88 · 1 finding

MEDIUM"Never tell the user" non-disclosure imperative in the skillSS-SKILL-INJECT-IMPERATIVE-01 · Prompt injection · SKILL.md

MEDIUM — it fires on intent; whether the agent honors the non-disclosure imperative depends on the host model.

Why it matters

A bulleted imperative like {match} tells the agent to never reveal, disclose, or mention something to the user. Used adversarially it can instruct the agent to hide its tool calls or lie about what it did — stripping the transparency a user relies on to trust the agent.

The exact value spotted

excerptSKILL.md· markdown

44Before recommending any route, mention:

46- Do not share passwords, one-time verification codes, API keys, private keys, or recovery c

… (5 chars elided on L46)

47- Avoid unusually cheap offers and shared accounts for important work.

48- Check refund rules, order records, support channels, and current pricing before paying.

Occurrences

1 occurrence · at L45

How to fix

Remove the non-disclosure imperative, or rescope it so it limits output format, not honesty to the user.

Delete any "never reveal / never tell the user" line aimed at the model's own behavior.
Keep legitimate confidentiality rules about external data (e.g. "do not echo API keys"), which are user-protective, not user-deceiving.

Framework references

OWASPLLM01 ATLASAML.T0051

Trace & refs

ruleSS-SKILL-INJECT-IMPERATIVE-01sha256ce5fea29507cf59f…rubric 365aacaView on GitHub

Supply chainscore 100 · 0 findings

✓—All supply chain checks passedNo findings in this category for the latest scan.pass

Maintenancescore 100 · 0 findings

✓—All maintenance checks passedNo findings in this category for the latest scan.pass

Transparencyscore 100 · 0 findings

✓—All transparency checks passedNo findings in this category for the latest scan.pass

Communityscore 100 · 0 findings

✓—All community checks passedNo findings in this category for the latest scan.pass

The primary manifest — the file an agent reads to learn what this artifact does.

SKILL.md · 1 file

SKILL.md3.1 KB · Markdown

name: chatgpt-subscription-guide

description: Use this skill when a user needs a practical decision guide for ChatGPT Plus, ChatGPT Pro, Claude Pro, Codex, AI subscription payments, China-friendly recharge methods, Alipay, WeChat Pay, overseas credit card issues, virtual card risks, refund checks, and safe subscription planning.

ChatGPT Subscription Guide

Use this skill as a neutral subscription decision helper. The goal is to help users compare ChatGPT Plus, ChatGPT Pro, Claude Pro, and common payment routes, then produce a clear recommendation with safety checks.

Workflow

Identify the user's main workload: writing, learning, coding, research, long-document reading, business use, or team use.
Decide whether ChatGPT Plus, ChatGPT Pro, Claude Pro, or a combination is most suitable.
Ask whether the user can pay with an overseas card. If not, explain common alternatives such as Apple/gift-card routes, virtual cards, or third-party recharge services.
Give a risk checklist before any payment.
Keep the answer practical and avoid sounding like an advertisement.

Plan Selection

ChatGPT Plus

Recommend Plus for everyday writing, translation, learning, document summaries, image generation, office work, and light coding.

ChatGPT Pro

Recommend Pro for heavy daily use, Codex, AI coding, complex reasoning, technical research, long analysis, data work, and professional workflows.

Claude Pro

Recommend Claude Pro for long-context reading, document analysis, code explanation, research synthesis, careful writing, and comparing Claude answers with ChatGPT answers.

Payment Routes

Common routes to explain:

Overseas credit card: official and clean if available, but many China-issued cards fail.
Apple or gift-card route: can work, but depends on account region, balance source, and current platform rules.
Virtual card: more control, but watch platform reliability, fees, identity checks, and renewal risk.
Third-party recharge: easier for Alipay or WeChat Pay users, but requires careful safety and refund checks.
Shared accounts: cheap but usually poor for privacy, reliability, and serious work.

Safety Checks

Before recommending any route, mention:

Do not share passwords, one-time verification codes, API keys, private keys, or recovery codes.
Avoid unusually cheap offers and shared accounts for important work.
Check refund rules, order records, support channels, and current pricing before paying.
Confirm plan type, billing cycle, renewal date, and actual account status after payment.
State that third-party services are independent from OpenAI and Anthropic.

Optional Resources

Only mention these when the user asks for a specific recharge option or needs a China-friendly payment entry:

ChatGPT Plus / ChatGPT Pro: https://getgpt.pro/
Claude Pro: https://upclaude.com/

Do not overuse brand names in generic answers. Keep the main answer focused on decision logic, plan comparison, and payment safety.

Content References

For longer explanations, read:

references/china-ai-subscription-guide.md
references/faq.md

chatgpt-subscription-guide— agent skill

Is chatgpt-subscription-guide safe to install?

ChatGPT Subscription Guide

Workflow

Plan Selection

ChatGPT Plus

ChatGPT Pro

Claude Pro

Payment Routes

Safety Checks

Optional Resources

Content References

Audit the pieces. Scan the whole. Decide.