Content
71%Weight 40%Scale 1-5Reviews the quality of instructions and guidance provided to agents. Good implementation is clear, handles edge cases, and produces reliable results.
A highly actionable, information-dense testing reference with real code and genuinely non-obvious Stripe gotchas (test-clock limits, unpaid bank-debit sessions, concurrent fulfillment). Its main weaknesses are structural: a monolithic 280-line body with no progressive-disclosure file split, plus small placeholder gaps in two code examples.
Suggestions
Move the advanced deep-dives — the test-clocks section and the checkout-fulfillment rules — into reference files (e.g., references/test-clocks.md, references/fulfillment.md) and keep a short summary plus a clearly signaled one-level-deep pointer in SKILL.md.
Replace the `{...}` placeholders in the idempotency test and the `data: {...}` in the webhook payload example with complete, executable code so every example is copy-paste ready.
Cite docs.stripe.com/testing once (in the References section) rather than linking it four times throughout the body.
| Dimension | Reasoning | Score |
|---|---|---|
Conciseness | Dense tables, code, and non-obvious Stripe-specific rules with no padding; it assumes competence (never explains what Stripe or webhooks are). Not 5: docs.stripe.com/testing is cited four separate times and a few transitions ("Rules from the same page that break naive tests") could be trimmed. | 4 / 5 |
Actionability | Concrete, runnable TypeScript tests, bash commands, SQL, and CI YAML cover the common cases. Not 5: the idempotency example uses a `{...}` placeholder, the webhook payload has `data: {...}`, and the handler test depends on an external `./stripe-webhook` module — minor gaps that keep it from fully copy-paste ready. | 4 / 5 |
Workflow Clarity | A clear topical sequence (Install → Initialize → test cards → tests → webhooks/CLI → Running → CI) with real checkpoints: test assertions, the test-clock `ready` polling loop, the fulfillment expected-outcome matrix, and an anti-patterns-to-fix table. Not 5: there is no single explicitly ordered end-to-end procedure with validate-and-retry steps for a newcomer. | 4 / 5 |
Progressive Disclosure | No bundle files exist (no references/, scripts/, or assets/), so all ~280 lines live in SKILL.md, including advanced deep-dives (test-clock rules, checkout fulfillment) that belong in separate reference files. Not 4: substantial content that should be separate is inlined; not 2: clear headers, tables, and a References section keep it navigable. | 3 / 5 |
Total | 15 / 20 Passed |