Getting it into your agent
One page per mod, every tool's command on it. A separate URL per tool would split the same page into five that compete with each other.
npx skills add adaptyteam/adapty-skills --skill purchase-testinggit clone --depth 1 https://github.com/adaptyteam/adapty-skillsWrote this? Show the measurements
A badge with what this costs and how it scanned, read live from this page, so it follows the numbers instead of freezing them. Markdown for a README, HTML for a documentation site or a project page.
[](https://agentmods.dev/skills/adaptyteam/adapty-skills/purchase-testing)<a href="https://agentmods.dev/skills/adaptyteam/adapty-skills/purchase-testing"><img src="https://agentmods.dev/badge/skills/adaptyteam/adapty-skills/purchase-testing/github.svg" alt="Measured on agentmods" height="20"></a>Or the 80×15 button, for a site that already has a row of RSS and ATOM ones. Only the verdict fits; the numbers stay here.
<a href="https://agentmods.dev/skills/adaptyteam/adapty-skills/purchase-testing"><img src="https://agentmods.dev/badge/skills/adaptyteam/adapty-skills/purchase-testing.svg" alt="Reviewed on agentmods" width="80" height="20"></a>What it costs to keep this loaded
Counted locally with the o200k_base tokenizer, which is exact for GPT models; Claude uses its own tokenizer and its counts differ. Treat this as one consistent yardstick across the catalogue rather than a bill. Prices are per million input tokens.
| Model | Per session | Once invoked |
|---|---|---|
| Fable 5.1 | $0.00128 | $0.06385 |
| Opus 5.5 | $0.00051 | $0.02554 |
| Sonnet 5.5 | $0.00026 | $0.01277 |
| Haiku 4.5 | $0.00013 | $0.00639 |
Grade B, and why
purchase-testing scanned grade B with 1 finding against 26 rules in 11 categories — prompt injection, anti-refusal, data exfiltration, privilege escalation, supply chain, agent snooping, system-prompt leakage, SSRF and excessive agency — measured 4d ago.
A static scan of the body, not an audit. Every finding is printed with the line that produced it so you can judge whether it matters here. A mod is markdown that instructs an agent; that is exactly why what it instructs is worth reading.
Subtle steeringmediumPrompt injection
Instructions that bias recommendations or shape behaviour without the user noticing.
alone as if it were the fix. While it is missing, it is the problem: never tell the user How it starts
The opening of the file, as written. The whole thing — 430 lines — stays where its author put it; the contents beside it link to each section on GitHub.
purchase-testing
What this answers
Why doesn't this purchase work — and what do I do about it? Also the forward case:
someone whose code builds and whose stores are connected, about to run their first
sandbox purchase — often straight after adapty-integration hands over.
The value is the order. A purchase failure can live in five different places, and the expensive ones involve a human clicking through a store console. So this skill checks everything it can reach itself, first, and only then asks anyone to open a browser.
This skill does not connect stores and does not write code. Creating store products,
uploading an App Store In-App Purchase Key, setting up a Google Play service account and
RTDN, and writing the SDK calls all belong to adapty-integration. When a check here
lands on one of those, offer the outcome ("I can add the code that closes the paywall
after a purchase. Want me to?") and invoke adapty-integration on a yes.
How you sound
Every message, not only the fixed blocks:
- A patient colleague who knows in-app purchases. The user is often testing a purchase for the first time. Assume they may know nothing about sandbox accounts or store consoles; when they ask, answer plainly and use their own app as the example. Friendly through patience, never through filler. Someone chasing a broken purchase is already frustrated, so the next action comes before any explanation.
- "I" for what you do, "you" for what they do. Lead with the answer; end on the one thing they need to do next. No opener ("I checked…"), no recap, no "let me know if…".
- Plain words over raw values. Show a value only when the user will type it or see
it elsewhere: a placement ID, a product ID, a menu path, a URL. Never field names
(
is_active), never "layer 2", never skill or tool names, never the paths of files you saved while working. - Bold at most one thing per paragraph. A label the user must pick or match (a test card, a menu item) goes in quotes, not bold.
- Full, clickable links to the exact page.
- Could not check it? Say how they can — where to click and what they should see.
Do not know something? Look it up before saying so — the CLI, the code, the
adapty-docsskill. - Offer the outcome, never the skill that will do it.
- Reply in the user's language. Menu names and buttons stay exactly as the phone or the console shows them.
What ships with it
2 files beside SKILL.md in the same directory: the scripts, references and assets a skill reads on demand. Not counted in the per-session cost; read them before you install if any of them is executable.
What this file has done since we first saw it
Hashed on every crawl. A supply-chain change to an agent config is a question of when, not whether, so the history is kept rather than the latest state alone.
- 4d ago Changed · +71 lines a485beffa2d1
- 28d ago Changed · +160 lines scan A → B 0b8f64c333f3
- 1mo ago Changed · +8 lines 31a0780d4487
- 1mo ago First seen · 191 lines · 128 tokens per session scan A d9f81f93eea6
purchase-testing is a skill published in the GitHub repository adaptyteam/adapty-skills (6 stars, last pushed today), licensed MIT. It adds 128 tokens to every session and 6,385 once invoked, about $0.0005 per session on Opus 5.5. A static security scan graded it B with 1 finding (subtle steering). No closer match exists in the catalogue, so it is treated as the original; first seen 2026-09-17.
Other skills, from other repositories
qa
QA-test a website or web app and return a 1-5 quality score (5 = flawless, 1 = broken) with evidence. Use when the user wants to test, QA, evaluate, score, or "check how good" a site, page, flow, or app — including a local dev server (e.g. "qa test localhost:5173", "does the checkout work?", "rate this landing page").…
payment-testing
Test payment and checkout flows end to end against PSP sandboxes — Stripe first, with the general pattern for Adyen/Braintree/PayPal. Covers Stripe test-mode card numbers and their decline codes, the 3DS/SCA challenge flow and its nested-iframe handling in Playwright, test clocks for subscription/billing-cycle…
postgres-commit-history-article
A workflow for comparing PostgreSQL source-code changes between two Git commits and, when useful, writing a Chinese Markdown article about them. PostgreSQL is a database system, and a commit is a saved change in a Git repository.
setup-trtmc-environment
Prepare a TensorRT-Model-Connect development or validation environment from a fresh checkout on a host without a known working setup.
synthetic-monitoring
AI-powered synthetic monitoring skill for e-commerce websites. Designs automated user journey tests for add-to-cart, checkout, and payment flows with alerting rules and performance baselines.
visual-regression-testing
AI-powered visual regression testing skill for e-commerce websites. Designs screenshot comparison workflows, mobile/desktop visual checks, and change detection alerts to prevent conversion-killing UI bugs.