feat: cap per-user AI token spend
CI and Deploy / test (push) Successful in 2m43s
CI and Deploy / deploy (push) Successful in 57s

This commit is contained in:
cesnimda
2026-07-30 23:32:13 +02:00
parent 09fc2b03f7
commit 792814b04e
5 changed files with 18 additions and 9 deletions
+1 -1
View File
@@ -197,7 +197,7 @@ Goal: commercialise. Last, per the guide's "do not over-engineer before needed.
| 7.6 | ✅ **DONE (2026-07-30)** — public CV (`/cv/{guid}`), privacy-first random links, revoke/rotate sharing | **P3** | **M** | 3.4, 4.2 | Anonymous rendering is isolated behind an explicit public flag, served with `noindex`, and revoked links cannot be restored accidentally. |
| 7.7 | ✅ **DONE (2026-07-30)** — three free CV themes plus five Premium themes, enforced by account entitlement and clearly locked in the picker | **P3** | **S** | 4.3, 7.2 | Existing Premium-theme CVs remain editable and exportable after downgrade so user data is never held hostage. |
| 7.8 | **DONE (2026-07-30)** — CI runs NuGet transitive vulnerability reporting and a production-only npm audit. The npm audit reports the existing no-fix advisory baseline without blocking unrelated deploys. | **P2** | **S** | none | Vulnerable dependencies are now visible before deployment. |
| 7.9 | **Per-user AI provider cost controls** | **P3** | **S** | 5.2, 7.2 | With `AI_PROVIDER=gemini` the "advanced AI" tier spends real money per call. Metering (5.2) measures; this enforces. |
| 7.9 | **DONE (2026-07-30)** — per-user monthly AI token ceilings (100k free, 1M Premium/Admin) enforced alongside generation limits and exposed in usage totals | **P3** | **S** | 5.2, 7.2 | Existing metering is the single accounting source; paid-provider spend now has both request and token ceilings. |
---