muse-agent-example/harness/test_skill_harness.py
zizi c9f69d9d6d 治理: Skill 测试治理第一阶段——harness 控制平面 + 实现测试迁出运行时目录
范围(不含 design-story-foundation、docs/、humanization/README.md 等进行中改动):

1. 新增 harness/ 控制平面
   - skill_harness.py 静态审计:32 个运行时 Skill 的 frontmatter/manifest/文档污染,当前 0 问题
   - run_selected.py 选择性执行器:manifest 与磁盘一一对账、依赖阻断、
     空跑与 skip-only 失败关闭、AST 测试形状门
   - manifests/skills.json:32 个 Skill 的合同责任方与协作领域登记
   - manifests/test-inventory.json:81 个测试资产登记
   - specs/skill-testing.md 与 README.md:测试分层、证据边界与 harness 职责

2. 实现测试从 .claude/skills/*/scripts/ 迁至 tests/skills/<skill>/
   - 71 个测试文件迁移并修复项目根与临时目录运行导入
   - 数据库触发器测试宽泛异常收窄为 psycopg.errors.RaiseException
   - 抽取离线大测试拆出真实 PG smoke(默认阻断,不计入离线通过)
   - 抽取 presence 去重边界拆出独立测试:493 + 78 = 571 项检查不变

3. 运行时文档清理
   - 13 个 SKILL.md 移除自测/离线验证段落、测试命令与测试文件事实源表述,
     只保留运行时合同;业务运行合同、额度、授权与离线模式均保留

4. SoT 同步
   - AGENTS.md:新增 Skill 领域索引(7 个合同责任方分组,覆盖 32 个运行时 Skill)
   - 领域 07:测试入口改由 harness/manifests/ 登记,SKILL.md 不承载测试命令
   - humanization 覆盖矩阵:活动测试路径同步迁移

验证证据: harness 自测 15 项 + runner 自测 13 项通过;静态审计 32 Skill / 0 问题;
73 个非数据库测试通过;8 个集成条目中 6 个 PostgreSQL 项被依赖门明确阻断;
py_compile 与 git diff --check 通过。未连接 PostgreSQL、网络、真实模型或额度。

已知边界: 真正 skill_behavior_eval 仍为 0,尚未验证任何 Skill 自然语言行为;
evaluate-frozen-replay 的 raw 存储边界冲突留待单独治理。
2026-08-19 01:50:20 +08:00

421 lines
15 KiB
Python
Raw Blame History

This file contains ambiguous Unicode characters

This file contains Unicode characters that might be confused with other characters. If you think that this is intentional, you can safely ignore this warning. Use the Escape button to reveal them.

#!/usr/bin/env python3
"""skill_harness 的纯标准库离线回归测试。
所有夹具都在临时目录中构造,不读取当前仓库的 Skill,也不依赖数据库、网络或模型。
"""
from __future__ import annotations
import io
import json
import tempfile
import unittest
from contextlib import redirect_stdout
from pathlib import Path
from typing import Optional
try:
from .skill_harness import audit_skills, main
except ImportError: # 允许直接执行 `.venv/bin/python harness/test_skill_harness.py`
from skill_harness import audit_skills, main
class SkillHarnessTests(unittest.TestCase):
def make_skill(
self,
root: Path,
directory: str,
*,
name: Optional[str] = None,
body: str = "# 合同\n\n只描述运行时行为。\n",
) -> Path:
skill_dir = root / ".claude" / "skills" / directory
skill_dir.mkdir(parents=True, exist_ok=True)
skill_path = skill_dir / "SKILL.md"
frontmatter_name = directory if name is None else name
skill_path.write_text(
f"---\nname: {frontmatter_name}\ndescription: 离线夹具\n---\n{body}",
encoding="utf-8",
)
self.write_manifest(root)
return skill_path
def write_manifest(
self,
root: Path,
entries: Optional[list[dict[str, object]]] = None,
*,
path: Optional[Path] = None,
) -> Path:
if entries is None:
entries = []
skills_root = root / ".claude" / "skills"
if skills_root.exists():
for directory in sorted(skills_root.iterdir(), key=lambda item: item.name):
if directory.is_dir():
entries.append(
{
"name": directory.name,
"contract_owner": "测试夹具",
"collaborates_with": [],
"skill_path": (
Path(".claude")
/ "skills"
/ directory.name
/ "SKILL.md"
).as_posix(),
}
)
manifest_path = path or root / "harness" / "manifests" / "skills.json"
manifest_path.parent.mkdir(parents=True, exist_ok=True)
manifest_path.write_text(
json.dumps(
{"schema_version": 1, "skills": entries},
ensure_ascii=False,
indent=2,
)
+ "\n",
encoding="utf-8",
)
return manifest_path
def test_passing_project_is_independent_of_current_repository(self) -> None:
with tempfile.TemporaryDirectory() as temporary:
root = Path(temporary)
self.make_skill(root, "alpha")
report = audit_skills(root)
self.assertTrue(report["ok"])
self.assertEqual(report["status"], "passed")
self.assertEqual(report["skills_scanned"], 1)
self.assertEqual(report["issues"], [])
def test_pollution_is_reported_for_each_obvious_category(self) -> None:
with tempfile.TemporaryDirectory() as temporary:
root = Path(temporary)
self.make_skill(
root,
"polluted",
body=(
"## 自测\n"
"pytest -q\n"
"import unittest\n"
"run test_sample.py and sample_test.py\n"
"覆盖率达到 100%。\n"
"测试通过。\n"
),
)
report = audit_skills(root)
codes = {issue["code"] for issue in report["issues"]}
self.assertFalse(report["ok"])
self.assertTrue(
{
"development_test_heading",
"pytest_reference",
"unittest_reference",
"test_file_reference",
"coverage_reference",
"test_pass_declaration",
}.issubset(codes)
)
def test_development_headings_support_multiple_levels(self) -> None:
with tempfile.TemporaryDirectory() as temporary:
root = Path(temporary)
self.make_skill(
root,
"nested-development-sections",
body=(
"### 自测\n"
"这里不能登记测试命令。\n"
"## 离线验证\n"
"这里也不能登记测试命令。\n"
),
)
report = audit_skills(root)
headings = [
issue
for issue in report["issues"]
if issue["code"] == "development_test_heading"
]
self.assertFalse(report["ok"])
self.assertEqual(len(headings), 2)
self.assertEqual([issue["line"] for issue in headings], [5, 7])
def test_check_contract_command_requires_development_section_context(self) -> None:
with tempfile.TemporaryDirectory() as temporary:
root = Path(temporary)
self.make_skill(
root,
"contextual-contract-command",
body=(
"业务合同字段名可以写 check_contract.py。\n"
"## 离线验证\n"
"python check_contract.py\n"
"## 业务说明\n"
"普通业务 check_contract.py 不是测试入口。\n"
),
)
report = audit_skills(root)
command_issues = [
issue
for issue in report["issues"]
if issue["code"] == "development_test_command_reference"
]
self.assertFalse(report["ok"])
self.assertEqual(len(command_issues), 1)
self.assertEqual(command_issues[0]["line"], 7)
def test_business_coverage_terms_are_not_development_pollution(self) -> None:
with tempfile.TemporaryDirectory() as temporary:
root = Path(temporary)
self.make_skill(
root,
"business-coverage",
body=(
"细纲覆盖率达到 100%。\n"
"事件覆盖率为 80%。\n"
"硬约束覆盖率 100%。\n"
),
)
report = audit_skills(root)
self.assertTrue(report["ok"])
self.assertNotIn(
"coverage_reference",
{issue["code"] for issue in report["issues"]},
)
def test_development_coverage_terms_are_reported(self) -> None:
with tempfile.TemporaryDirectory() as temporary:
root = Path(temporary)
self.make_skill(
root,
"development-coverage",
body=(
"coverage.py\n"
"pytest-cov\n"
"pytest --cov=harness\n"
"coverage report\n"
"测试覆盖率达到 90%。\n"
"覆盖率达到 80%。\n"
),
)
report = audit_skills(root)
self.assertFalse(report["ok"])
self.assertIn(
"coverage_reference",
{issue["code"] for issue in report["issues"]},
)
def test_frontmatter_name_must_match_skill_directory(self) -> None:
with tempfile.TemporaryDirectory() as temporary:
root = Path(temporary)
self.make_skill(root, "alpha", name="beta")
report = audit_skills(root)
self.assertFalse(report["ok"])
self.assertIn(
"name_directory_mismatch",
{issue["code"] for issue in report["issues"]},
)
def test_duplicate_frontmatter_names_are_reported(self) -> None:
with tempfile.TemporaryDirectory() as temporary:
root = Path(temporary)
self.make_skill(root, "alpha", name="shared")
self.make_skill(root, "beta", name="shared")
report = audit_skills(root)
duplicates = [
issue for issue in report["issues"] if issue["code"] == "duplicate_name"
]
self.assertFalse(report["ok"])
self.assertEqual(len(duplicates), 1)
self.assertEqual(duplicates[0]["name"], "shared")
self.assertEqual(len(duplicates[0]["paths"]), 2)
def test_clean_manifest_can_be_selected_explicitly(self) -> None:
with tempfile.TemporaryDirectory() as temporary:
root = Path(temporary)
self.make_skill(root, "alpha")
manifest_path = self.write_manifest(
root,
path=root / "clean-manifest.json",
)
output = io.StringIO()
with redirect_stdout(output):
return_code = main(
[
"--root",
str(root),
"--manifest",
str(manifest_path),
"--json",
"--quiet",
]
)
payload = json.loads(output.getvalue())
self.assertEqual(return_code, 0)
self.assertTrue(payload["ok"])
self.assertTrue(payload["manifest"]["loaded"])
self.assertEqual(payload["manifest"]["path"], "clean-manifest.json")
def test_error_manifest_fails_with_structured_issues(self) -> None:
with tempfile.TemporaryDirectory() as temporary:
root = Path(temporary)
self.make_skill(root, "alpha")
self.make_skill(root, "beta")
broken_manifest = self.write_manifest(
root,
entries=[
{
"name": "alpha",
"contract_owner": "",
"collaborates_with": ["上下文与知识检索", 3],
"skill_path": ".claude/skills/beta/SKILL.md",
},
{
"name": "alpha",
"contract_owner": "测试夹具",
"collaborates_with": [],
"skill_path": ".claude/skills/beta/SKILL.md",
},
{
"name": "extra",
"contract_owner": "测试夹具",
"collaborates_with": [],
"skill_path": ".claude/skills/extra/SKILL.md",
},
],
path=root / "broken-manifest.json",
)
output = io.StringIO()
with redirect_stdout(output):
return_code = main(
[
"--root",
str(root),
"--manifest",
str(broken_manifest),
"--json",
]
)
payload = json.loads(output.getvalue())
codes = {issue["code"] for issue in payload["issues"]}
self.assertEqual(return_code, 1)
self.assertFalse(payload["ok"])
self.assertTrue(all(isinstance(issue, dict) for issue in payload["issues"]))
self.assertTrue(
{
"manifest_contract_owner_invalid",
"manifest_collaborates_with_invalid",
"manifest_duplicate_skill_path",
"manifest_duplicate_name",
"manifest_skill_missing",
"manifest_skill_extra",
"manifest_skill_path_mismatch",
"manifest_name_directory_mismatch",
"manifest_skill_path_missing",
}.issubset(codes)
)
def test_missing_manifest_fails_closed(self) -> None:
with tempfile.TemporaryDirectory() as temporary:
root = Path(temporary)
self.make_skill(root, "alpha")
(root / "harness" / "manifests" / "skills.json").unlink()
report = audit_skills(root)
self.assertFalse(report["ok"])
self.assertIn(
"manifest_missing",
{issue["code"] for issue in report["issues"]},
)
def test_missing_skills_directory_fails_closed(self) -> None:
with tempfile.TemporaryDirectory() as temporary:
root = Path(temporary)
report = audit_skills(root)
self.assertFalse(report["ok"])
self.assertIn(
"skills_directory_missing",
{issue["code"] for issue in report["issues"]},
)
def test_skill_directory_without_skill_file_fails(self) -> None:
with tempfile.TemporaryDirectory() as temporary:
root = Path(temporary)
(root / ".claude" / "skills" / "missing").mkdir(parents=True)
report = audit_skills(root)
self.assertFalse(report["ok"])
self.assertIn(
"missing_skill_file",
{issue["code"] for issue in report["issues"]},
)
def test_empty_skill_file_is_reported(self) -> None:
with tempfile.TemporaryDirectory() as temporary:
root = Path(temporary)
skill_path = self.make_skill(root, "empty")
skill_path.write_text("", encoding="utf-8")
report = audit_skills(root)
self.assertFalse(report["ok"])
self.assertIn(
"empty_skill_file",
{issue["code"] for issue in report["issues"]},
)
def test_json_cli_and_exit_code(self) -> None:
with tempfile.TemporaryDirectory() as temporary:
root = Path(temporary)
self.make_skill(root, "alpha")
output = io.StringIO()
with redirect_stdout(output):
return_code = main(["--root", str(root), "--json", "--quiet"])
payload = json.loads(output.getvalue())
self.assertEqual(return_code, 0)
self.assertTrue(payload["ok"])
self.assertEqual(payload["skills_scanned"], 1)
(root / ".claude" / "skills" / "alpha" / "SKILL.md").write_text(
"---\nname: alpha\n---\n## 测试\n",
encoding="utf-8",
)
output = io.StringIO()
with redirect_stdout(output):
return_code = main(["--root", str(root), "--json"])
failed_payload = json.loads(output.getvalue())
self.assertEqual(return_code, 1)
self.assertFalse(failed_payload["ok"])
if __name__ == "__main__":
unittest.main(verbosity=2)