diff --git a/Makefile b/Makefile
index 4562048..c2d6274 100644
--- a/Makefile
+++ b/Makefile
@@ -30,6 +30,10 @@ fmt-check:
 lint:
 	golangci-lint run $(GO_PKGS)
 
-# Run the test suite.
+# Run the test suite. Quiet on success; on failure, rerun verbosely for the
+# full output and still fail the target (the first run already proved the
+# tests are broken, so a flaky pass on the rerun must not rescue the build).
 test:
-	go test $(GO_PKGS)
+	@go test -timeout 30s -race -cover $(GO_PKGS) || \
+		{ echo "--- Rerunning with -v for details ---"; \
+		  go test -timeout 30s -race -v $(GO_PKGS); exit 1; }
diff --git a/TODO.md b/TODO.md
index e99d21b..c0833bc 100644
--- a/TODO.md
+++ b/TODO.md
@@ -34,6 +34,19 @@ wizard commands).
 
 # Completed Steps
 
+- 2026-08-09 Policy-shaped `make test` (`make-test-policy-pattern`): the `test:`
+  target was a bare `go test $(GO_PKGS)` and now runs
+  `-timeout 30s -race -cover` with the mandated conditional verbose rerun (on
+  failure it reruns with `-v` and then `exit 1`, so a flaky pass on the second
+  attempt cannot rescue the build). `$(GO_PKGS)` is kept rather than hardcoding
+  `./...`. The substance was `-race`, not the Makefile edit: this is the first
+  time the suite has run under the race detector, and it is clean — no data
+  races across five consecutive uncached runs, including the tcell terminal
+  layer and the `os.Exit`-path playthrough tests. Wall clock 5.1s cold
+  (including the race build) and ~2.3s warm, against the 20s policy budget. The
+  failure path was exercised with a throwaway failing test to confirm the rerun
+  fires and `make` exits non-zero. Build tooling only; no game behavior change.
+
 - 2026-08-07 Canonical linter config (`golangci-v2.12.2`): replaced
   `.golangci.yml` with the shared canonical config (v2 schema; settings now live
   under `linters.settings`, so the `lll`/`funlen`/`cyclop`/`dupl` thresholds