@@ -164,17 +164,20 @@ jobs:
       # the build on any skip.
       # The API reads PGDATABASE (api/config.py). The schema lives in
       # te_mgmt_dev, not the job-level default "postgres" (issue #45).
+      # Accessibility scans need a browser and the frontend, not this database;
+      # they run in the `accessibility` job below (#168) and are reported here as
+      # NOT RUN, not as skipped.
       - name: Full test suite — final result with skip accounting
         env:
           PGDATABASE: te_mgmt_dev
-        run: python3 scripts/test_report.py --strict
+        run: python3 scripts/test_report.py --markers "not accessibility" --strict
 
       # #194: measure api/services coverage with per-loader floors.
       - name: api/services coverage gate
         env:
           PGDATABASE: te_mgmt_dev
         run: |
-          python -m pytest tests/ \
+          python -m pytest tests/ -m "not accessibility" \
             --cov=api/services \
             --cov-report=json:coverage-services.json \
             --cov-report=term-missing \
@@ -189,3 +192,62 @@ jobs:
           path: evals/reports/
           if-no-files-found: ignore
           retention-days: 30
+
+  # #168: WCAG 2.1 A and AA scans of the frontend (csv-table-hub-main) with
+  # Playwright + axe-core in Chromium, against the Vite dev server. The API is
+  # stubbed, so this job needs no database and a backend-only change cannot
+  # alter what it scans. Known violations are recorded in
+  # tests/accessibility/known_violations.json: the job fails on a NEW violation
+  # and when a recorded one is fixed (so the list only shrinks). Not yet a
+  # required check; make it one after a few green runs (see the issue).
+  accessibility:
+    name: accessibility (Playwright + axe-core, WCAG 2.1 AA)
+    runs-on: ubuntu-latest
+    steps:
+      - name: Checkout
+        uses: actions/checkout@v7
+
+      - name: Setup Python
+        uses: actions/setup-python@v7
+        with:
+          python-version: '3.11'
+          cache: 'pip'
+          cache-dependency-path: |
+            requirements-dev.txt
+            requirements-a11y.txt
+            api/requirements.txt
+
+      - name: Setup Node
+        uses: actions/setup-node@v6
+        with:
+          node-version: '22'
+          cache: 'npm'
+          cache-dependency-path: csv-table-hub-main/package-lock.json
+
+      - name: Install dev dependencies
+        # api/requirements.txt is needed too: tests/conftest.py imports the API
+        # app, so pytest collection fails without FastAPI.
+        run: pip install -r requirements-dev.txt -r api/requirements.txt -r requirements-a11y.txt
+
+      - name: Install frontend dependencies
+        working-directory: csv-table-hub-main
+        run: npm ci
+
+      - name: Install Chromium for Playwright
+        run: python -m playwright install --with-deps chromium
+
+      # Prints the same final block as the other jobs. --strict fails on any skip.
+      # The scans start `npm run dev` themselves (tests/accessibility/conftest.py).
+      - name: Accessibility scans — final result with skip accounting
+        env:
+          A11Y_REPORT_DIR: accessibility-report
+        run: python3 scripts/test_report.py --markers accessibility --strict
+
+      - name: Upload accessibility report
+        if: always()
+        uses: actions/upload-artifact@v7
+        with:
+          name: accessibility-report
+          path: accessibility-report/
+          if-no-files-found: ignore
+          retention-days: 30
@@ -1,4 +1,5 @@
 .PHONY: test-free test-gate test-evals test-e2e test-all \
+        test-unit test-system test-sit test-performance test-accessibility test-uat \
         lint lint-diff health \
         eval-list eval-compare eval-summary select-tests \
         csv-load csv-list csv-demo
@@ -26,12 +27,42 @@ test-evals:
 test-e2e:
 	pytest -m "e2e or integration or parity" tests/ -v
 
-# All: every test tier including evals.
+# All: every test tier including evals. Accessibility scans need a browser and the
+# frontend, so they run on their own: `make test-accessibility` (tests/accessibility/README.md).
 # Mirrors: bun run test:evals:all
 test-all:
-	pytest tests/ --cov=build/csv --cov=evals --cov-report=term-missing
+	pytest tests/ -m "not accessibility" --cov=build/csv --cov=evals --cov-report=term-missing
 	python3 evals/runner.py --tiers p,i,s,e --verbose
 
+# ── One category at a time (#167) ─────────────────────────────────────────────
+# tests/ is split by what a test needs: unit (nothing), system (the API and
+# PostgreSQL), system_integration (whole pipeline, all environments), then
+# performance, accessibility and uat. See tests/README.md.
+
+# pytest exits 5 when a folder holds no tests yet; that is not a failure for a
+# category that has none so far.
+define run_category
+	python3 -m pytest $(1) -v; rc=$$?; if [ $$rc -eq 5 ]; then echo "No tests in $(1) yet."; exit 0; fi; exit $$rc
+endef
+
+test-unit:
+	$(call run_category,tests/unit)
+
+test-system:
+	$(call run_category,tests/system)
+
+test-sit:
+	$(call run_category,tests/system_integration)
+
+test-performance:
+	$(call run_category,tests/performance)
+
+test-accessibility:
+	$(call run_category,tests/accessibility)
+
+test-uat:
+	$(call run_category,tests/uat)
+
 # ── Code quality ──────────────────────────────────────────────────────────────
 
 # Full flake8 + bandit scan over all Python source.
 
@@ -244,8 +244,8 @@ Interactive docs: `http://localhost:8000/docs`
 python -m pytest tests\ -m "unit or regression or security or snapshot" -q
 
 # Integration tests — requires running API + provisioned database
-python -m pytest tests\test_api_integration.py -v
-python -m pytest tests\test_te_loader_integration.py -v
+python -m pytest tests\system\test_api_integration.py -v
+python -m pytest tests\system\test_te_loader_integration.py -v
 
 # Per-environment contract tests
 python -m pytest tests\environments\ -v
@@ -338,7 +338,7 @@ organisations ──< personnel
 | Connection Limit | 10 | 15 | 25 | 50 |
 | Seed Data | ✅ | ✅ | ❌ | ❌ |
 
-Staging and Prod are **never seeded**. The `tests/environments/` contract tests
+Staging and Prod are **never seeded**. The `tests/system_integration/environments/` contract tests
 verify this automatically.
 
 ---
 
@@ -78,7 +78,7 @@ async def tenant_scope(authorization: str = Header(default=""),
 
 
 # M2c (#94): which roles may perform each action. docs/ROLES.md shows the same
-# table, and tests/test_roles.py checks that the two agree.
+# table, and tests/system/test_roles.py checks that the two agree.
 PERMISSIONS: dict[str, frozenset[str]] = {
     "preview":     frozenset({"uploader", "tenant_admin"}),
     "upload":      frozenset({"uploader", "tenant_admin"}),
 
@@ -5,7 +5,7 @@
 O-04 update: the table is no longer delete-only. Successful uploads also write a
 row, through the shared writer in api/services/audit.py, so this feed now covers
 creations as well as destructions. Integration coverage lives in
-tests/test_api_integration.py::AuditLogTrail."""
+tests/system/test_api_integration.py::AuditLogTrail."""
 
 from __future__ import annotations
 
 
@@ -237,8 +237,8 @@ Per the repository's no-skip policy, the integration group **fails** with
 remediation text when the database is unreachable rather than skipping.
 
 ```bash
-python -m pytest tests/test_api_unit.py -m unit      # no database
-python -m pytest tests/test_api_unit.py tests/test_api_integration.py  # both files
+python -m pytest tests/unit/test_api_unit.py -m unit      # no database
+python -m pytest tests/unit/test_api_unit.py tests/system/test_api_integration.py  # both files
 python scripts/test_report.py --strict                # whole suite
 ```
 
@@ -254,8 +254,8 @@ with remediation text rather than skipping, per the no-skip policy.
 
 ## Also required
 
-CI installs only `requirements-dev.txt`. Because `tests/test_api_unit.py` and
-`tests/test_api_integration.py` import FastAPI at module level, both
+CI installs only `requirements-dev.txt`. Because `tests/unit/test_api_unit.py` and
+`tests/system/test_api_integration.py` import FastAPI at module level, both
 workflows need:
 
 ```yaml
 
@@ -50,15 +50,15 @@ The three layers can break independently, so we keep them physically separate. T
 | `tests/run_tests.sh` | Bash wrapper that sources `config.local.env` |
 | `tests/run_python_tests.ps1` | Windows runner — invoked by the GitHub Actions workflow |
 | `tests/conftest.py` | pytest env-var isolation between tests |
-| `tests/test_csv_validator.py` | unittest for `build/csv/validator.py` |
-| `tests/test_csv_utilise.py` | unit tests for `build/csv_utilise.sh` argument parsing |
-| `tests/test_csv_loader_arbitrary_shapes.py` | integration: arbitrary CSV shapes through loader → PG (skips without PG) |
-| `tests/test_e2e_pipeline.py` | e2e: CSV → validate → load → verify (DB half skips without PG) |
-| `tests/test_parity.py` | cross-environment row-count / schema parity (skips without PG) |
-| `tests/test_regression.py` | pinned tests for previously found bug classes |
-| `tests/test_security.py` | static credential/SQL-pattern scans |
-| `tests/test_snapshot.py` | golden-file output comparisons (`tests/snapshots/`) |
-| `tests/test_evals_runner.py` | unittest for `evals/runner.py` itself |
+| `tests/unit/test_csv_validator.py` | unittest for `build/csv/validator.py` |
+| `tests/unit/test_csv_utilise.py` | unit tests for `build/csv_utilise.sh` argument parsing |
+| `tests/system_integration/test_csv_loader_arbitrary_shapes.py` | integration: arbitrary CSV shapes through loader → PG (skips without PG) |
+| `tests/system_integration/test_e2e_pipeline.py` | e2e: CSV → validate → load → verify (DB half skips without PG) |
+| `tests/system_integration/test_parity.py` | cross-environment row-count / schema parity (skips without PG) |
+| `tests/unit/test_regression.py` | pinned tests for previously found bug classes |
+| `tests/unit/test_security.py` | static credential/SQL-pattern scans |
+| `tests/unit/test_snapshot.py` | golden-file output comparisons (`tests/snapshots/`) |
+| `tests/unit/test_evals_runner.py` | unittest for `evals/runner.py` itself |
 
 ### `evals/` — data-driven scenarios
 
@@ -110,7 +110,7 @@ These are left at the repo root and excluded from the layer model. They can move
 ## Path conventions inside the layers
 
 - Code inside `evals/runner.py` computes `PROJECT_ROOT = Path(__file__).resolve().parent.parent` and reaches into `build/csv/validator.py`, `build/environments/env_dev.sql`, `tests/run_all_tests.sql` from there.
-- Code inside `tests/test_csv_validator.py` does `Path(__file__).resolve().parents[1] / "build" / "csv" / "validator.py"`.
+- Code inside `tests/unit/test_csv_validator.py` does `Path(__file__).resolve().parents[2] / "build" / "csv" / "validator.py"`.
 - Shell scripts inside `build/` cd into their own directory and use relative paths (`./adapters/...`, `./schema/...`).
 - Nothing in `build/` references `tests/` or `evals/`.
 
 
@@ -96,7 +96,7 @@ temp_documents   -> references test_programs   (independent of the chain above)
 Then, in the same commit, move each table from `known_missing_tables` into
 `expected_tables` in
 `evals/expected/tier_x/01_sqlite_schema_structure.json`. Tier X will fail until
-you do — `tests/test_evals_tier_e_x.py` asserts that the two lists together
+you do — `tests/unit/test_evals_tier_e_x.py` asserts that the two lists together
 account for all 12 tables, and that everything listed as missing is genuinely
 absent. That is the reminder working as designed.
 
 
@@ -69,7 +69,7 @@ the environment SQL templates. Copy `build/config.env.example` to
 
 > **Staging and prod must never be seeded.** The `SEED_STAGING` and `SEED_PROD`
 > variables default to `false`. Changing them to `true` will insert test data
-> into production. The `tests/test_parity.py` contract tests verify this.
+> into production. The `tests/system_integration/test_parity.py` contract tests verify this.
 
 ### Additional build-script variables
 
 
@@ -117,9 +117,9 @@ scenario still fails on drift in either direction. It also executes the DDL
 rather than `adapter_sqlite.sh`, so the adapter scripts themselves remain
 unexercised — a smaller, separate gap.
 
-Also mitigated: `tests/test_parity.py::TestAllEnvironmentsHaveRequiredTables`
+Also mitigated: `tests/system_integration/test_parity.py::TestAllEnvironmentsHaveRequiredTables`
 runs against all four PostgreSQL environments, and
-`tests/test_evals_tier_e_x.py` unit-tests both tiers' comparison logic with
+`tests/unit/test_evals_tier_e_x.py` unit-tests both tiers' comparison logic with
 negative controls for added columns, changed types, changed nullability,
 reordered columns, and missing or extra tables.