fix(ci): wait for ordinary eval agents within job timeout

.github/workflows/eval-pr-run.yml

@@ -284,6 +284,7 @@ jobs:
 
   run-evals:
     name: Run PR Evals
+    timeout-minutes: 90
     needs: [discover, gate]
     if: >-
       !cancelled() &&
@@ -359,6 +360,8 @@ jobs:
           ANTHROPIC_DEFAULT_OPUS_MODEL: "claude-opus-4-6"
           ANTHROPIC_MODEL: "claude-opus-4-6"
           CLAUDE_CODE_SUBPROCESS_ENV_SCRUB: "1"
+          # Wait for eval agents to finish; the job timeout bounds total runtime.
+          CLAUDE_CODE_PRINT_BG_WAIT_CEILING_MS: "0"
         run: |
           IFS=',' read -ra SKILLS <<< "$SKILLS_CSV"
           failed=0

plugins/sdlc-workflow/scripts/test_native_fullsend_eval_ci.py

@@ -515,6 +515,9 @@ def run_ordinary_eval_step(tmp_path, mode, skills="first"):
 workspace = Path(re.search(r"Workspace: (.+)", args[args.index("-p") + 1])[1])
 with open("calls.jsonl", "a") as capture:
     capture.write(json.dumps(args) + "\n")
+with open("wait-settings.jsonl", "a") as capture:
+    capture.write(json.dumps({name: os.environ.get(name) for name in (
+        "CLAUDE_CODE_PRINT_BG_WAIT_CEILING_MS", "CLAUDE_CODE_SUBPROCESS_ENV_SCRUB")}) + "\n")
 mode = os.environ["EVAL_STUB_MODE"] if workspace.name == "first-eval-pr" else "correct"
 if mode == "exit":
     sys.exit(7)
@@ -538,16 +541,37 @@ def run_ordinary_eval_step(tmp_path, mode, skills="first"):
         source = tmp_path / "pr-head/evals" / skill
         source.mkdir(parents=True)
         (source / "evals.json").write_text('{"evals": [{"id": 1}]}')
-    script = script_step("run-evals", "Run PR evals")["run"].replace(
+    step = script_step("run-evals", "Run PR evals")
+    script = step["run"].replace(
         'workspace="/tmp/${skill}-eval-pr"', f'workspace="{tmp_path}/${{skill}}-eval-pr"')
+    env = dict(os.environ, PATH=f"{tools}:{os.environ['PATH']}",
+               SKILLS_CSV=skills, EVAL_STUB_MODE=mode)
+    env.pop("CLAUDE_CODE_PRINT_BG_WAIT_CEILING_MS", None)
+    for name in ("CLAUDE_CODE_PRINT_BG_WAIT_CEILING_MS", "CLAUDE_CODE_SUBPROCESS_ENV_SCRUB"):
+        if name in step["env"]:
+            env[name] = step["env"][name]
     result = subprocess.run(["bash", "-e", "-o", "pipefail", "-c", script], cwd=tmp_path,
-                            env=dict(os.environ, PATH=f"{tools}:{os.environ['PATH']}",
-                                     SKILLS_CSV=skills, EVAL_STUB_MODE=mode),
+                            env=env,
                             capture_output=True, text=True)
     calls = [json.loads(line) for line in (tmp_path / "calls.jsonl").read_text().splitlines()]
     return result, calls
 
 
+def test_ordinary_eval_passes_background_wait_setting_to_each_cli(tmp_path):
+    """Both real shell invocations inherit the workflow wait policy and scrubbing."""
+    result, calls = run_ordinary_eval_step(tmp_path, "correct", "first,second")
+    assert result.returncode == 0, result.stdout + result.stderr
+    assert len(calls) == 2
+    settings = [json.loads(line) for line in (tmp_path / "wait-settings.jsonl").read_text().splitlines()]
+    assert settings == [{"CLAUDE_CODE_PRINT_BG_WAIT_CEILING_MS": "0",
+                         "CLAUDE_CODE_SUBPROCESS_ENV_SCRUB": "1"}] * 2
+
+
+def test_ordinary_eval_job_bounds_background_wait():
+    """Disabling the CLI idle ceiling still leaves an explicit total CI limit."""
+    assert workflow()["jobs"]["run-evals"].get("timeout-minutes") == 90
+
+
 @pytest.mark.parametrize("mode", ["absent", "misplaced", "missing-benchmark.json",
                                  "missing-feedback.json", "missing-summary.md",
                                  "empty-benchmark.json", "empty-feedback.json", "empty-summary.md",