From d2b3cb01c0e49f83863c466b92b356158b53faa2 Mon Sep 17 00:00:00 2001 From: Johannes du Plessis Date: Fri, 5 Jun 2026 10:31:00 -0700 Subject: [PATCH] fix: flag skipped CI tests in reviewer (#1428) Co-authored-by: open-swe[bot] --- agent/reviewer.py | 7 ++++++- tests/test_reviewer.py | 2 ++ 2 files changed, 8 insertions(+), 1 deletion(-) diff --git a/agent/reviewer.py b/agent/reviewer.py index e834373c..79beb9dc 100644 --- a/agent/reviewer.py +++ b/agent/reviewer.py @@ -189,7 +189,12 @@ carefully before reaching for unchanged code. HTML/template rendering, or cross-origin behavior, trace the resolution path. Don't just suggest tidying — confirm what actually happens on the hit, miss, and error paths. -5. **Verify library / framework usage you're not certain of.** If a +5. **CI/CD test enforcement.** When the diff touches workflow files, build + scripts, package scripts, Makefiles, test runner config, or CI-specific + conditionals, check whether any test suite is no longer run in CI/CD. + Specifically flag tests being skipped, disabled, removed, made non-blocking, + or conditionally bypassed without an equivalent replacement. +6. **Verify library / framework usage you're not certain of.** If a stdlib, ORM, or framework call's semantics matter to the change, confirm the contract before assuming a bug or assuming safety. diff --git a/tests/test_reviewer.py b/tests/test_reviewer.py index 881ebc4c..f4285af3 100644 --- a/tests/test_reviewer.py +++ b/tests/test_reviewer.py @@ -17,6 +17,8 @@ def test_reviewer_system_prompt_formats_without_keyerror() -> None: ) assert "acme/repo" in prompt assert "The bar" in prompt + assert "CI/CD test enforcement" in prompt + assert "Specifically flag tests being skipped" in prompt assert "benchmark" not in prompt.lower() assert "golden" not in prompt.lower() assert "at least 1 finding" not in prompt.lower()