testing strategy

---

title: Testing Strategy created: 2026-07-09 updated: 2026-07-15 type: concept tags: [infrastructure] confidence: high source_files:

• docs/TESTING.md
• Makefile
• integrity-dashboard/playwright.config.ts
• integrity-dashboard/e2e/global-setup.ts
• integrity-dashboard/e2e/global-teardown.ts
• integrity-dashboard/e2e/constants.ts

---

The full picture — per-package unit/integration tests, integrity-dashboard's component tests, and the new real-browser E2E layer — lives in docs/TESTING.md; this page is the wiki's pointer into it plus the facts worth having in the knowledge graph directly.

Table of contents

• Three layers
• Ground rule
• Convention going forward

Three layers

flowchart TB
    L1["Layer 1 — Per-package (make test)<br/>forge / nargo / cargo / pytest / vitest,<br/>real toolchains, no mocked DB or chain"]
    L2["Layer 2 — Playwright E2E (make test-e2e)<br/>real Chromium against a freshly-booted<br/>local stack (anvil + oracle + Postgres + Redis)"]
    L3["Layer 3 — Hosted CI [NOT BUILT]<br/>no git remote yet — make test / make test-e2e<br/>run manually by a human or agent"]

    L1 --> L2 --> L3

Loading

1. Per-package (make test): forge test (contracts, 165), nargo test (integrity-zkp), cargo test (integrity-oracle, 54 backend + scoring-core lib + real e2e), pytest (integrity-sdk 97, integrity-cli 57, bcc_middleware 75+28 OPA, integrity-userapi 33 against a real Postgres), npm test/vitest (integrity-dashboard, real components + msw-mocked HTTP boundary).
2. Playwright E2E (make test-e2e, new 2026-07-09): a real Chromium browser against the real integrity-dashboard app, talking to a real, freshly-booted local stack (integrity-dashboard/e2e/global-setup.ts: a dedicated local anvil + real genesis/market deploy, ephemeral Postgres+Redis, a real integrity-oracle instance, one real seed agent registered through the real SDK flow). Local anvil only — never live Base Sepolia, for speed/cost/determinism (mirrors the convention Integrity SDK's and Integrity Oracle's own test suites already use).
3. Not yet built: hosted CI. This repo has no git remote today, so there's nothing for GitHub Actions to trigger from — make test/make test-e2e run by a human or agent are the enforcement mechanism until that changes (see .agents/AGENTS.md §6). An honest, documented gap.

Ground rule

No silent mocks, at every layer. msw in integrity-dashboard's vitest suite is the one deliberate mock in the whole pyramid, scoped to exactly the HTTP boundary — the Playwright layer exists specifically to prove the real request/response actually works, which the mocked layer can't.

Convention going forward

As integrity-dashboard's dashboard pages get built out (task #21 — Markets, Leaderboard, Wallet, Capital Allocation, Cognition, Identity, Health, Landing), each page's Playwright spec ships in the same pass as the page, not backfilled later.