fix(cli): repair installed preflight command

Author: reacher-z

CHANGELOG.md

@@ -4,7 +4,22 @@ All notable changes to RewardHarness are recorded here. Versions follow [SemVer]
 
 ## [Unreleased]
 
-Future changes after v0.2.1.
+Future changes after v0.2.2.
+
+## [0.2.2] — 2026-08-18
+
+### Fixed
+
+- Fixed the installed `rewardharness check` command reparsing the process-wide
+  arguments after the main CLI had already consumed the `check` subcommand.
+  The preflight now receives explicit arguments and works consistently through
+  both `rewardharness check` and `python scripts/check_env.py`.
+
+### Added
+
+- Added `--endpoints` and `--timeout` to `rewardharness check`, with validation
+  for non-positive timeouts and direct forwarding to endpoint probes.
+- Added regression coverage for CLI-to-diagnostics option forwarding.
 
 ## [0.2.1] — 2026-08-18
 
@@ -162,6 +177,7 @@ deprecated `src` namespace remains as a compatibility layer for v0.2.
 - `make demo` and `make benchmark` default to `--library-dir examples/seed_library` for non-empty starting state.
 - `make help` is now a credentials matrix showing what each target actually needs.
 
+[0.2.2]: https://github.com/TIGER-AI-Lab/RewardHarness/releases/tag/v0.2.2
 [0.2.1]: https://github.com/TIGER-AI-Lab/RewardHarness/releases/tag/v0.2.1
 [0.2.0]: https://github.com/TIGER-AI-Lab/RewardHarness/releases/tag/v0.2.0
 [0.1.2]: https://github.com/TIGER-AI-Lab/RewardHarness/releases/tag/v0.1.2

CITATION.cff

@@ -1,7 +1,7 @@
 cff-version: 1.2.0
 message: "If you use this software, please cite the paper."
 title: "RewardHarness: Self-Evolving Agentic Post-Training"
-version: "0.2.1"
+version: "0.2.2"
 date-released: "2026-08-17"
 type: software
 authors:

README.md

@@ -43,6 +43,8 @@ Read [`WALKTHROUGH.md`](WALKTHROUGH.md) for the 9-step path from `git clone` to
 
 ## Updates
 
+- **2026-08-18** — `v0.2.2`: fixed the installed `rewardharness check`
+  command and added explicit endpoint path and timeout controls.
 - **2026-08-18** — `v0.2.1`: deterministic wheel/sdist content auditing,
   stricter release metadata verification, and safer clean-build publishing.
 - **2026-08-17** — `v0.2.0`: stable PyPI release with verified wheel/sdist,

SECURITY.md

@@ -23,7 +23,7 @@ Only the current `main` branch and the most recent tagged release receive securi
 | Version | Supported |
 |---|---|
 | `main` | ✅ |
-| `v0.2.1` (latest) | ✅ |
+| `v0.2.2` (latest) | ✅ |
 | earlier | ❌ |
 
 ## Disclosure history

TROUBLESHOOTING.md

@@ -51,7 +51,7 @@ export GEMINI_LOCATION="global"  # or us-central1 / europe-west4
 Quick check:
 
 ```bash
-python -c "from src.gemini_client import get_client; get_client(); print('OK')"
+python -c "from rewardharness.clients.gemini import get_client; get_client(); print('OK')"
 ```
 
 **`PermissionDenied: 403 Vertex AI API has not been used in project ... or it is disabled`**

rewardharness/_version.py

@@ -1,3 +1,3 @@
 """Single source of truth for the RewardHarness package version."""
 
-__version__ = "0.2.1"
+__version__ = "0.2.2"

rewardharness/cli.py

@@ -38,7 +38,9 @@ def build_parser() -> argparse.ArgumentParser:
     parser.add_argument("--version", action="version", version=f"%(prog)s {__version__}")
     subparsers = parser.add_subparsers(dest="command", required=True)
 
-    subparsers.add_parser("check", help="validate local credentials and endpoints")
+    check_parser = subparsers.add_parser("check", help="validate local credentials and endpoints")
+    check_parser.add_argument("--endpoints", default=str(default_endpoints_path()))
+    check_parser.add_argument("--timeout", type=float, default=3.0)
     subparsers.add_parser("release-status", help="show canonical package and release identifiers")
 
     inspect_parser = subparsers.add_parser("inspect", help="inspect a Library registry")
@@ -137,7 +139,7 @@ def main(argv: Sequence[str] | None = None) -> int:
     if args.command == "check":
         from rewardharness.diagnostics import main as diagnostics_main
 
-        return diagnostics_main()
+        return diagnostics_main(["--endpoints", args.endpoints, "--timeout", str(args.timeout)])
     if args.command == "release-status":
         print(json.dumps(ReleaseIdentity.current().to_dict(), indent=2, sort_keys=True))
         return 0

rewardharness/diagnostics.py

@@ -27,6 +27,7 @@
 import sys
 import urllib.error
 import urllib.request
+from collections.abc import Sequence
 from concurrent.futures import ThreadPoolExecutor
 from pathlib import Path
 
@@ -131,7 +132,7 @@ def check_endpoints(endpoints_path: str, timeout: float = 3.0) -> bool:
         print(f"  [{WARN}] no endpoints listed in {endpoints_path}")
         return True
     # Import the canonical constant so this script's "expected" stays in
-    # lockstep with src.sub_agent.SUBAGENT_MODEL (already env-var-aware).
+    # lockstep with rewardharness.evaluation.engine.SUBAGENT_MODEL.
     # Resolved lazily so check_env.py keeps working even if src/ isn't on
     # the path (e.g. running this from outside the repo root).
     try:
@@ -163,14 +164,22 @@ def check_endpoints(endpoints_path: str, timeout: float = 3.0) -> bool:
     return True  # endpoint probe is informational only
 
 
-def main():
+def main(argv: Sequence[str] | None = None) -> int:
     parser = argparse.ArgumentParser(description=__doc__.split("\n")[0])
     parser.add_argument(
         "--endpoints",
         default="configs/endpoints.txt",
         help="Path to endpoints file (default: configs/endpoints.txt)",
     )
-    args = parser.parse_args()
+    parser.add_argument(
+        "--timeout",
+        type=float,
+        default=3.0,
+        help="Per-endpoint timeout in seconds (default: 3.0)",
+    )
+    args = parser.parse_args(argv)
+    if args.timeout <= 0:
+        parser.error("--timeout must be positive")
 
     print("RewardHarness preflight check\n")
 
@@ -183,7 +192,7 @@ def main():
     print("\n4. Service-account credentials")
     ok_creds = check_credentials_file() if ok_env else False
     print("\n5. vLLM endpoints (informational)")
-    check_endpoints(args.endpoints)
+    check_endpoints(args.endpoints, timeout=args.timeout)
 
     print()
     all_required = ok_py and ok_imports and ok_env and ok_creds

tests/test_check_env.py

@@ -9,6 +9,7 @@
 from unittest.mock import patch
 
 import rewardharness.diagnostics as diagnostics
+from rewardharness.cli import main as cli_main
 
 
 def _load_check_env():
@@ -62,3 +63,31 @@ def test_non_200_returns_status_line(self):
             _url, err, served = mod._probe_one("http://x/v1", timeout=1.0)
         assert "HTTP 500" in err
         assert served == ""
+
+
+def test_cli_forwards_check_options_without_reparsing_process_argv(monkeypatch):
+    received = []
+
+    def fake_main(argv=None):
+        received.append(argv)
+        return 0
+
+    monkeypatch.setattr(diagnostics, "main", fake_main)
+    assert cli_main(["check", "--endpoints", "custom.txt", "--timeout", "0.25"]) == 0
+    assert received == [["--endpoints", "custom.txt", "--timeout", "0.25"]]
+
+
+def test_diagnostics_main_uses_explicit_options(monkeypatch):
+    observed = []
+    monkeypatch.setattr(diagnostics, "check_python_version", lambda: True)
+    monkeypatch.setattr(diagnostics, "check_imports", lambda: True)
+    monkeypatch.setattr(diagnostics, "check_env_vars", lambda: True)
+    monkeypatch.setattr(diagnostics, "check_credentials_file", lambda: True)
+    monkeypatch.setattr(
+        diagnostics,
+        "check_endpoints",
+        lambda path, timeout: observed.append((path, timeout)) or True,
+    )
+
+    assert diagnostics.main(["--endpoints", "custom.txt", "--timeout", "0.25"]) == 0
+    assert observed == [("custom.txt", 0.25)]

tests/test_infrastructure.py

@@ -97,7 +97,7 @@ def test_cli_reports_package_version(capsys):
     with pytest.raises(SystemExit) as raised:
         build_parser().parse_args(["--version"])
     assert raised.value.code == 0
-    assert capsys.readouterr().out.strip() == "rewardharness 0.2.1"
+    assert capsys.readouterr().out.strip() == "rewardharness 0.2.2"
 
 
 def test_gemini_text_and_candidate_fallback(monkeypatch):