Freeze pricing lineage at archived shared-core v1.3.0; refresh metadata wording; bump dashboard Next to 16.3.1 and vitest to 4 (audit clean)

Author: FishRaposo

AGENTS.md

@@ -42,7 +42,7 @@ llm-gateway/
 make install      # npm ci
 make dev          # tsx watch src/index.ts
 make build        # tsc -> dist/
-make test         # vitest run  -> 150 passing (backend); dashboard has 27 more
+make test         # vitest run  -> 149 passing (backend); dashboard has 27 more
 make lint         # tsc --noEmit + eslint
 make typecheck    # tsc --noEmit
 make docker-up    # redis + gateway
@@ -51,17 +51,17 @@ make demo         # build + print a sample OpenAI-compatible request
 
 ## Current State
 
-**Functional, migrated, green, hardened.** `tsc --noEmit` clean; **150 backend vitest tests
-pass** (19 files), plus **27 dashboard tests** (3 files); `eslint .` clean; `next build` green.
+**Functional, migrated, green, hardened.** `tsc --noEmit` clean; **149 backend vitest tests
+pass** (20 files), plus **27 dashboard tests** (3 files); `eslint .` clean; `next build` green.
 Storage degrades gracefully: when the native `better-sqlite3` binding is unavailable (e.g. no
 C++ build tools), the API-key/audit stores fall back to in-memory — the full test suite runs
 without compilation.
 
 The root vitest suite is scoped to `tests/` via `vitest.config.ts`; the dashboard has its own
-jsdom vitest project under `dashboard/`. `src/shared/pricing.ts` now mirrors
-`shared_core.pricing` per-1M rates (`MODEL_PRICING_PER_1M`, single source of truth, sync
-documented in its header) and derives the per-token catalog from it; parity is enforced by
-`tests/pricing.test.ts`. The optional Next.js dashboard gained a demo-mode fallback, an
+jsdom vitest project under `dashboard/`. `src/shared/pricing.ts` carries the per-1M
+rates (`MODEL_PRICING_PER_1M`) mirrored from `shared_core.pricing` — the lineage is
+**frozen at the archived shared-core v1.3.0** (2026-08-13), the table is self-owned, and
+the frozen parity snapshot is pinned by `tests/pricing.test.ts`. The optional Next.js dashboard gained a demo-mode fallback, an
 `ErrorBoundary`, extracted testable helpers, component tests, and an optional Playwright smoke
 spec.
 
@@ -70,9 +70,9 @@ spec.
 - ✅ Audit-log column schema and Prometheus label keys (`llm_gateway_*`) documented as a
   superset of the Python `llm-cost-latency-monitor` `LLMCall` cost-record and pinned by
   `tests/monitorAlignment.test.ts` so one dashboard reads both.
-- ✅ Pricing table mirrors `shared_core.pricing` (data parity, `MODEL_PRICING_PER_1M`); sync
-  procedure documented and parity-tested.
-- ⏭️ `claude-3-5-haiku` rate diverges from shared_core (1.0/5.0 vs 0.8/4.0); deferred because
+- ✅ Pricing table mirrors `shared_core.pricing` (data parity, `MODEL_PRICING_PER_1M`) —
+  lineage frozen at the archived v1.3.0; the table is self-owned and parity-tested.
+- ⏭️ `claude-3-5-haiku` rate diverges from the frozen shared-core snapshot (1.0/5.0 vs 0.8/4.0); deferred because
   changing it would move existing cost outputs (golden-gated). Pinned + tracked in roadmap.md.
 - ✅ Canonical-gateway question settled 2026-08-12: `knowledgeops` (incl. `services/llm-gateway`) was consolidated into groundtruth and archived on GitHub — this standalone gateway is the canonical one.
 

README.md

@@ -2,7 +2,7 @@
 
 [![Tests](https://img.shields.io/badge/tests-passing-brightgreen)]() [![Coverage](https://img.shields.io/badge/coverage-88%25-brightgreen)]() [![TypeScript](https://img.shields.io/badge/TypeScript-3178C6?logo=typescript)]() [![Express](https://img.shields.io/badge/Express-000?logo=express)]() [![Redis](https://img.shields.io/badge/Redis-DC382D?logo=redis)]()
 
-**Enterprise LLM proxy with fallback, caching, rate limiting, and budget enforcement.**
+**LLM proxy with routing, guardrails, cost control, and fallback.**
 
 Route requests across multiple providers (OpenAI, Anthropic, Gemini, Ollama) with automatic failover, response caching, and per-API-key cost control.
 
@@ -154,7 +154,7 @@ curl -X POST http://localhost:3000/v1/chat/completions \
 
 ## 9. Testing Strategy
 
-The backend suite has **150 vitest tests** across 19 files; the optional dashboard has **27**
+The backend suite has **149 vitest tests** across 20 files; the optional dashboard has **27**
 component/unit tests across 3 files. Run `npx vitest run` at the repo root and
 `cd dashboard && npx vitest run` for the UI.
 
@@ -167,7 +167,7 @@ tests/
 ├── proxy.test.ts            # Full request pipeline, error handling, streaming
 ├── rateLimit.test.ts        # Sliding-window limiting, per-key config, 429 path
 ├── logging.test.ts          # Structured audit logging + API-key redaction
-├── pricing.test.ts          # Pricing parity with shared_core.pricing (golden-gated)
+├── pricing.test.ts          # Frozen pricing parity snapshot (archived shared_core v1.3.0, golden-gated)
 ├── costService.test.ts      # Cost calc + model-pricing normalization
 └── monitorAlignment.test.ts # Audit + Prometheus key parity with the Python monitor
 ```
@@ -176,8 +176,9 @@ tests/
 - **Integration tests**: full pipeline with the mock provider (no real API calls).
 - **Error path tests**: provider failures, budget exceeded, policy denied, rate limited.
 - **Golden-output tests**: pricing/cost values are pinned so refactors can't silently move them.
-- **Cross-language tests**: pin the pricing rates, audit cost-record columns, and Prometheus
-  key names shared with `shared_core.pricing` and the Python `llm-cost-latency-monitor`.
+- **Cross-language tests**: pin the frozen pricing lineage (archived `shared_core.pricing`
+  v1.3.0) plus the audit cost-record columns and Prometheus key names shared with the
+  Python `llm-cost-latency-monitor` (now consolidated into agenttrace).
 - **Dashboard tests** (`dashboard/`): pure data helpers, the `ErrorBoundary`, and the
   console in demo / live / empty states.
 

dashboard/next-env.d.ts

@@ -1,5 +1,7 @@
 /// <reference types="next" />
 /// <reference types="next/image-types/global" />
+import "./.next/types/routes.d.ts";
+import "./.next/types/root-params.d.ts";
 
 // NOTE: This file should not be edited
-// see https://nextjs.org/docs/basic-features/typescript for more information.
+// see https://nextjs.org/docs/app/api-reference/config/typescript for more information.