Trim to BCApps review checklists + bump version to 0.7.0

Drop the inline-instructions prompt-gating scaffolding (config.yaml block,
prompt._category_context CODE_REVIEW context, and its test). That wiring stays
on experiment branches; main only carries the 6 domain checklist md files as
dormant assets plus the benchmark version bump.

Author: wenjiefan

pyproject.toml

@@ -4,7 +4,7 @@ build-backend = "setuptools.build_meta"
 
 [project]
 name = "bcbench"
-version = "0.6.0"
+version = "0.7.0"
 description = "Benchmarking tool for Business Central (AL) ecosystem, inspired by SWE-Bench"
 readme = "README.md"
 requires-python = ">=3.13"

src/bcbench/agent/shared/config.yaml

@@ -65,16 +65,6 @@ prompt:
       - body: concise description of the issue (string, required)
 
     If there are no findings, write an empty array. Write only valid JSON to review.json, with no surrounding markdown or commentary.
-    {% if inline_instructions_enabled %}
-    Evaluate the diff against every one of the following domain checklists. Read each file under `.github/instructions/` (or `.claude/instructions/`) and apply its guidance:
-      - `instructions/security.md` — security
-      - `instructions/performance.md` — performance
-      - `instructions/privacy.md` — privacy and data protection
-      - `instructions/style.md` — AL style and conventions
-      - `instructions/accessibility.md` — accessibility
-      - `instructions/upgrade.md` — upgrade and data-migration safety
-    Only raise findings supported by one of these checklists. For each finding, cite the concrete rule it violates and point to the exact file and line in the diff.
-    {% endif %}
 
 # controls:
 # 1. whether to copy custom instructions from `src/bcbench/agent/shared/instructions/<sanitized-repo>/`

src/bcbench/agent/shared/prompt.py

@@ -18,28 +18,21 @@ def _transform_image_paths(content: str) -> str:
 def build_prompt(entry: BaseDatasetEntry, repo_path: Path, config: dict, category: EvaluationCategory, al_mcp: bool = False) -> str:
     prompt_config = config.get("prompt", {})
     template_str = prompt_config.get(f"{category.value}-template")
-
-    context = {
-        "repo_path": repo_path,
-        "task": _transform_image_paths(entry.get_task()),
-        "project_paths": ", ".join(entry.project_paths),
-        "include_project_paths": prompt_config.get("include_project_paths"),
-        "al_mcp": al_mcp,
-    }
-    context |= _category_context(category, config)
-
-    return Template(template_str).render(**context)
-
-
-def _category_context(category: EvaluationCategory, config: dict) -> dict:
-    match category:
-        case EvaluationCategory.TEST_GENERATION:
-            mode = config.get("prompt", {}).get("test-generation-input", "problem-statement")
-            return {
-                "is_gold_patch": mode in ("gold-patch", "both"),
-                "is_problem_statement": mode in ("problem-statement", "both"),
-            }
-        case EvaluationCategory.CODE_REVIEW:
-            return {"inline_instructions_enabled": config.get("instructions", {}).get("enabled", False)}
-        case _:
-            return {}
+    include_project_paths = prompt_config.get("include_project_paths")
+
+    test_gen_input: str = prompt_config.get("test-generation-input", "problem-statement")
+    is_gold_patch: bool = category == EvaluationCategory.TEST_GENERATION and test_gen_input in ("gold-patch", "both")
+    is_problem_statement: bool = category == EvaluationCategory.TEST_GENERATION and test_gen_input in ("problem-statement", "both")
+
+    task = _transform_image_paths(entry.get_task())
+
+    template = Template(template_str)
+    return template.render(
+        repo_path=repo_path,
+        task=task,
+        project_paths=", ".join(entry.project_paths),
+        include_project_paths=include_project_paths,
+        is_gold_patch=is_gold_patch,  # only relevant for test-generation
+        is_problem_statement=is_problem_statement,  # only relevant for test-generation
+        al_mcp=al_mcp,  # whether AL MCP server is enabled
+    )

tests/test_copilot_prompt.py

@@ -130,18 +130,3 @@ def test_build_prompt_test_generation_both_mode(tmp_path: Path):
     assert "[HAS_PATCH]" in result  # gold patch should be indicated
     assert "[HAS_ISSUE]" in result  # problem statement should be indicated
     assert "Fix payment validation bug" in result  # task should be included in both mode
-
-
-def test_code_review_prompt_checklists_only_with_custom_instructions(tmp_path: Path):
-    entry = create_dataset_entry(instance_id="microsoftInternal__NAV-9", project_paths=["App/app"])
-    repo_path = tmp_path / "navapp"
-    repo_path.mkdir()
-    problem_dir = create_problem_statement_dir(tmp_path, "Review the change")
-    template = {"code-review-template": "/review\n{% if inline_instructions_enabled %}Read instructions/security.md{% endif %}"}
-
-    with patch.object(type(entry), "problem_statement_dir", property(lambda self: problem_dir)):
-        with_inline = build_prompt(entry, repo_path, {"prompt": template, "instructions": {"enabled": True}}, EvaluationCategory.CODE_REVIEW)
-        without_inline = build_prompt(entry, repo_path, {"prompt": template, "instructions": {"enabled": False}}, EvaluationCategory.CODE_REVIEW)
-
-    assert "security.md" in with_inline
-    assert "security.md" not in without_inline