Pricing copy: mock interviews billed by the minute, sessions stop at 0

[alpha5611331]

Closes #20
Backend: PowerInterviewAI/backend#67 · Client: PowerInterviewAI/client-app#138

What

Live and mock now share one rate, 10 credits a minute, and a session stops when the credits run out. A mock interview still delivers its report.

• src/lib/plans.ts: CREDIT_RATES becomes { perMinute: 10 }. mockSessionPrice is replaced by mockSessionMinutes, which mirrors the desktop app's estimate.
• Pricing section "What a session spends": the mock line is now per minute (an 8-question session is about 20 minutes, about 200 credits), and both lines say what happens at zero.
• Mock interview page: the pricing card changes from "Priced by the question" to "Priced by the minute".
• FAQ, en and ru: "What does a mock interview cost?" and "How do credits work?" are rewritten. The JSON-LD picks this up through src/config/faq.ts.
• Docs: mock-interview.md (new cost table by length), live-interview.md, introduction.md, how-it-works.md.
• CLAUDE.md: the pricing notes are updated.

Checks

• pnpm lint and tsc --noEmit are clean.
• next build was not run locally, because a dev server was running on this checkout (see the note about build clobbering dev). Please build before merging.

Release

Merge after the backend is deployed. The copy describes the new behaviour (stop at 0, per-minute mock).

🤖 Generated with Claude Code

[vercel]

The latest updates on your projects. Learn more about Vercel for GitHub.

Project	Deployment	Actions	Updated
power-interview-ai	Ready	Preview	Oct 6, 2026 3:24pm UTC

[gitar-bot]

Gitar is working

_Gitar

[alpha5611331]

Review notes: copy only, no logic beyond CREDIT_RATES/mockSessionMinutes. Every per-question price reference in src/ is gone (checked with grep, en and ru). The FAQ JSON-LD reads the same message files, so it stays in sync. Run next build before merging (it wasn't run locally, because a dev server was up). Merge only after backend#67 is deployed: the copy promises the stop at 0.

[alpha5611331]

Review pass 2: fixed f2a504d. The mock line in the pricing note hard-coded the minutes noun after the number, which is wrong Russian for most counts. It now goes through pluralize with minuteNoun, as the hero CLAUDE.md requires, and the Russian sentence uses the counting form ('примерно 20 минут и 200 кредитов'). I checked the rest of the copy against the backend and client: 10/min for both kinds of session, a 1-minute minimum to start, warnings at 5 and 1 minutes, the stop at 0 with the mock report still delivered, the live save prompt, and the length estimates (3, 5, 8, 12 questions = 8, 13, 20, 30 min, from the client's mockSessionMinutes). It all matches, and there's no per-question wording left in src/.