fix(config): preserve inherited prerelease readiness settings

[Haloukaidi]

Setting ARGUS_SKILL_ALLOW_BACKEND_PRERELEASE=1 is ignored by several readiness entry points because their default False is forwarded as an explicit rejection. The low-level readiness check only inherits the environment when the argument is None.

Preserve the omitted / false / true distinction through CLI parsing, Doctor, Web diagnostics, and interactive/non-interactive setup. An explicit False still overrides the environment, and the default policy remains unchanged when no opt-in is configured.

Validation on Windows / Python 3.13:

• 73 tests passed across tests/core/test_backend_readiness.py and tests/tools/test_setup_readiness.py.
• The 20 new regression cases cover omitted CLI options, Doctor inheritance, Web policy precedence with environment unset/0/1, and both setup paths.
• Ruff passed for the modified Python files; git diff --check passed.
• The regression cases are in the existing backend readiness module, already selected by Windows CI and included by the Linux suite. Remote CI has not yet run.

[Copilot]

🟢 Approval recommended

The behavioral changes align with the PR’s stated goal and are backed by targeted regression tests, with only minor test-code formatting maintainability nits noted.

Pull request overview

This PR fixes prerelease readiness inheritance by preserving a tri-state allow_prerelease value (None/False/True) across CLI parsing, Doctor context, Web diagnostics, and setup, so ARGUS_SKILL_ALLOW_BACKEND_PRERELEASE=1 is not accidentally overridden by forwarded default False.

Changes:

• Change allow_prerelease defaults from False to None in key readiness entry points so environment inheritance is preserved when the option is omitted.
• Stop coercing parsed CLI values via bool(...) so “omitted” remains distinguishable from an explicit policy.
• Add regression tests covering CLI parsing, Doctor inheritance, Web precedence behavior, and setup propagation.

File summaries

File	Description
tests/core/test_backend_readiness.py	Adds regression coverage for omitted vs explicit prerelease policy propagation across entry points.
argus_skill/webapi/diagnostics.py	Makes diagnostics’ prerelease policy tri-state by defaulting to None and forwarding through preflight.
argus_skill/tools/setup.py	Preserves omitted prerelease policy through setup flows by accepting/forwarding `bool
argus_skill/maintenance/doctor.py	Updates DoctorContext default to None so Doctor inherits env policy when not explicitly set.
argus_skill/apps/cli/_parser.py	Ensures --allow-prerelease parses as None when omitted (not an implicit False).
argus_skill/apps/cli/_core.py	Stops collapsing allow_prerelease to False when absent; forwards the parsed tri-state.

Review details

Suppressed comments (2)

tests/core/test_backend_readiness.py:703

• New test uses inconsistent spacing around assignment compared to the rest of the file, which reduces readability and makes future diffs noisier.

def test_cli_context_preserves_unspecified_value(tmp_path):
    args=SimpleNamespace(life_dir=str(tmp_path), resume="", allow_prerelease=None)
    assert _core._maintenance_context(args).allow_prerelease is None

tests/core/test_backend_readiness.py:710

• This test packs several statements onto single lines (including a long readiness call), which is inconsistent with surrounding tests and makes failures harder to diagnose; consider formatting it across lines.

def test_doctor_default_inherits_documented_environment(monkeypatch, tmp_path):
    fake_cli(monkeypatch)
    ctx=DoctorContext(global_root=tmp_path,project_root=tmp_path)
    report=readiness.check_backend_readiness("codex","subscription_cli",probe_auth=False,allow_prerelease=ctx.allow_prerelease)
    assert report.ok, report.problems

• Files reviewed: 6/6 changed files
• Comments generated: 1
• Review effort level: Lite

---

💡 Add a code-review agent skill or configure MCP servers for context-aware, tailored reviews. Learn more in the docs.