Run flat terrain scenario in budget report

Co-authored-by: Codex <codex@openai.com>

tools/airdef.py

@@ -157,6 +157,10 @@ def test_wasm():
         cwd=ROOT / "airdef-mod", env=build_env)
 
 
+BUDGET_SCENARIO = "tests/budgets/wasm_flat_budget.toml"
+BUDGET_ASSERTED_SHA256 = "bcc0359289ab20ab3de608bcc130f72d2cf88f9f6fbc470d7b6846e9db5d11df"
+
+
 def test_budgets():
     """Run the opt-in heavy scenario and always write a machine-readable report."""
     report_path = Path(os.environ.get("AIRDEF_BUDGET_REPORT") or
@@ -164,13 +168,15 @@ def test_budgets():
     report = {
         "schema": 1,
         "command": "test-budgets",
-        "scenario": "tests/budgets/wasm_budget.toml",
+        "scenario": BUDGET_SCENARIO,
+        "scenario_render_world": "flat at y=64",
         "status": "failed",
         "exit_code": None,
         "scenario_counts": {"scripts": 1, "steps": None, "raid_requests": None,
                              "raid_units_requested": None, "air_defense_units": None,
                              "layout_unit_limit": None, "frames": None, "ticks": None,
-                             "accepted_raids_observed": None, "peak_airborne_observed": None},
+                             "accepted_raids_asserted": None, "snapshot_airborne_asserted": None,
+                             "snapshot_tick": None, "peak_airborne_observed": None},
         "runtime_limits": {"source": "horizon/crates/horizon-mods/src/runtime.rs",
                            "metering": "source_configured; measured_per_scenario: unavailable"},
     }
@@ -205,7 +211,7 @@ def test_budgets():
         lock_text = (ROOT / "horizon" / "Cargo.lock").read_text()
         wasmtime = re.search(r'\[\[package\]\]\s*name = "wasmtime"\s*version = "([^"]+)"', lock_text)
         report["wasmtime_version"] = wasmtime.group(1) if wasmtime else None
-        fixture = ROOT / "airdef-mod" / "tests" / "budgets" / "wasm_budget.toml"
+        fixture = ROOT / "airdef-mod" / BUDGET_SCENARIO
         fixture_text = fixture.read_text()
         counts = report["scenario_counts"]
         counts["steps"] = len(re.findall(r"(?m)^\[\[step\]\]$", fixture_text))
@@ -238,7 +244,6 @@ def test_budgets():
                                ", ".join(k for k, ok in identity_checks.items() if not ok))
         build_env = dict(os.environ)
         build_env["PATH"] = str(Path(cargo()).parent) + os.pathsep + build_env.get("PATH", "")
-        fixture = ROOT / "airdef-mod" / "tests" / "budgets" / "wasm_budget.toml"
         with tempfile.TemporaryDirectory(prefix="airdef-wasm-budget-") as temp:
             isolated = Path(temp)
             (isolated / "tests").mkdir()
@@ -252,6 +257,11 @@ def test_budgets():
             report["exit_code"] = result.returncode
             report["status"] = "passed" if result.returncode == 0 else "failed"
             report["output"] = result.stdout[-12000:]
+            # These assertions belong to this exact fixture, not to arbitrary edited scripts.
+            if result.returncode == 0 and report["scenario_sha256"] == BUDGET_ASSERTED_SHA256:
+                counts["accepted_raids_asserted"] = 2
+                counts["snapshot_airborne_asserted"] = 184
+                counts["snapshot_tick"] = 2102
     except Exception as exc:
         report["error"] = f"{type(exc).__name__}: {exc}"
     except SystemExit as exc:

tools/test_budget_report.py

@@ -39,7 +39,7 @@ class BudgetReportFailureTests(unittest.TestCase):
             '[[package]]\nname = "wasmtime"\nversion = "48.0.7"\n'
         )
         (self.mod_root / "src/layout.rs").write_text("pub const MAX_UNITS: usize = 2000;\n")
-        (self.mod_root / "tests/budgets/wasm_budget.toml").write_text(
+        (self.mod_root / "tests/budgets/wasm_flat_budget.toml").write_text(
             '[[step]]\ncommand = "airdef test"\n'
         )
         self.artifact.write_bytes(b"minimal synthetic hmod")
@@ -95,7 +95,7 @@ class BudgetReportFailureTests(unittest.TestCase):
         return report
 
     def test_counts_are_requests_and_derived_from_fixture(self):
-        (self.mod_root / "tests/budgets/wasm_budget.toml").write_text(
+        (self.mod_root / "tests/budgets/wasm_flat_budget.toml").write_text(
             '[[step]]\ncommand = "airdef deploy s400"\n'
             '[[step]]\ncommand = "airdef wave kh101 80 4000"\n'
             '[[step]]\nticks = 450\nframes = 225\n'
@@ -106,7 +106,22 @@ class BudgetReportFailureTests(unittest.TestCase):
         self.assertEqual((counts["steps"], counts["air_defense_units"]), (4, 1))
         self.assertEqual((counts["raid_requests"], counts["raid_units_requested"]), (2, 120))
         self.assertEqual((counts["ticks"], counts["frames"]), (450, 225))
-        self.assertIsNone(counts["accepted_raids_observed"])
+        self.assertIsNone(counts["accepted_raids_asserted"])
+        self.assertIsNone(counts["peak_airborne_observed"])
+
+    def test_success_reports_only_assertions_of_exact_fixture(self):
+        source = Path(__file__).resolve().parent.parent / "airdef-mod" / airdef.BUDGET_SCENARIO
+        (self.mod_root / airdef.BUDGET_SCENARIO).write_bytes(source.read_bytes())
+        def backend(command, cwd, **kwargs):
+            self.assertEqual((Path(cwd) / "tests/wasm_budget.toml").read_bytes(), source.read_bytes())
+            return CompletedProcess(args=command, returncode=0, stdout="test passed")
+        with mock.patch.object(airdef, "mod"), mock.patch.object(airdef.subprocess, "run", side_effect=backend):
+            airdef.test_budgets()
+        report = self.read_report()
+        self.assertEqual(report["scenario"], airdef.BUDGET_SCENARIO)
+        self.assertEqual(report["scenario_render_world"], "flat at y=64")
+        counts = report["scenario_counts"]
+        self.assertEqual((counts["accepted_raids_asserted"], counts["snapshot_airborne_asserted"], counts["snapshot_tick"]), (2, 184, 2102))
         self.assertIsNone(counts["peak_airborne_observed"])
 
     def test_early_system_exit_writes_report_and_skips_backend(self):