ForcePilot/backend/package/yuxi/services/skill_service.py

from __future__ import annotations

import asyncio
import hashlib
import re
import shutil
import tempfile
import threading
import uuid
import zipfile
from pathlib import Path, PurePosixPath
from typing import Any

import yaml
from sqlalchemy.ext.asyncio import AsyncSession
from yuxi import config as sys_config
from yuxi.repositories.skill_repository import SkillRepository
from yuxi.services.mcp_service import get_enabled_mcp_server_names
from yuxi.storage.postgres.models_business import Skill
from yuxi.utils.logging_config import logger

SKILL_SLUG_PATTERN = re.compile(r"^[a-z0-9]+(-[a-z0-9]+)*$")
SKILL_NAME_PATTERN = SKILL_SLUG_PATTERN

TEXT_FILE_EXTENSIONS = {
    ".md",
    ".txt",
    ".py",
    ".js",
    ".ts",
    ".json",
    ".yaml",
    ".yml",
    ".toml",
    ".ini",
    ".cfg",
    ".conf",
    ".xml",
    ".html",
    ".css",
    ".sql",
    ".sh",
    ".bat",
    ".ps1",
    ".env",
    ".csv",
    ".tsv",
    ".rst",
    ".ipynb",
    ".vue",
    ".jsx",
    ".tsx",
}

BUILTIN_SKILL_OPERATOR = "builtin-system"
_THREAD_SKILLS_LOCK = threading.Lock()
_THREAD_SKILLS_LOCKS: dict[str, threading.Lock] = {}


class BuiltinSkillUpdateConflictError(ValueError):
    def __init__(self, message: str):
        super().__init__(message)
        self.needs_confirm = True


def _get_thread_skills_lock(thread_id: str) -> threading.Lock:
    with _THREAD_SKILLS_LOCK:
        lock = _THREAD_SKILLS_LOCKS.get(thread_id)
        if lock is None:
            lock = threading.Lock()
            _THREAD_SKILLS_LOCKS[thread_id] = lock
        return lock


def _normalize_string_list(values: list[str] | None) -> list[str]:
    if not values:
        return []
    normalized: list[str] = []
    seen: set[str] = set()
    for value in values:
        if not isinstance(value, str):
            continue
        item = value.strip()
        if not item or item in seen:
            continue
        seen.add(item)
        normalized.append(item)
    return normalized


def is_valid_skill_slug(slug: str) -> bool:
    if not isinstance(slug, str):
        return False
    return bool(SKILL_SLUG_PATTERN.match(slug.strip()))


def get_skills_root_dir() -> Path:
    root = Path(sys_config.save_dir) / "skills"
    root.mkdir(parents=True, exist_ok=True)
    return root


def get_thread_skills_root_dir(thread_id: str) -> Path:
    safe_thread_id = str(thread_id or "").strip()
    if not safe_thread_id:
        raise ValueError("thread_id is required")
    if not re.fullmatch(r"[A-Za-z0-9_-]+", safe_thread_id):
        raise ValueError("thread_id contains invalid characters")

    root = Path(sys_config.save_dir) / "threads" / safe_thread_id / "skills"
    root.mkdir(parents=True, exist_ok=True)
    return root


def sync_thread_visible_skills(thread_id: str, selected_slugs: list[str] | None) -> Path:
    skills_root = get_skills_root_dir().resolve()
    thread_skills_root = get_thread_skills_root_dir(thread_id)
    normalized_slugs = [slug for slug in _normalize_string_list(selected_slugs) if is_valid_skill_slug(slug)]
    visible_slugs = set(normalized_slugs)
    with _get_thread_skills_lock(thread_id):
        for entry in thread_skills_root.iterdir():
            if entry.name in visible_slugs:
                continue
            if entry.is_dir() and not entry.is_symlink():
                shutil.rmtree(entry)
            else:
                entry.unlink()

        for slug in normalized_slugs:
            source_dir = (skills_root / slug).resolve()
            target_dir = thread_skills_root / slug

            try:
                source_dir.relative_to(skills_root)
            except ValueError:
                continue
            if not source_dir.is_dir():
                if target_dir.exists() or target_dir.is_symlink():
                    if target_dir.is_dir() and not target_dir.is_symlink():
                        shutil.rmtree(target_dir)
                    else:
                        target_dir.unlink()
                continue

            if target_dir.exists():
                if target_dir.is_symlink():
                    target_dir.unlink()
                elif target_dir.is_dir():
                    if _dirs_equal(target_dir, source_dir):
                        continue
                    shutil.rmtree(target_dir)
                else:
                    target_dir.unlink()

            temp_target = thread_skills_root / f".{slug}.tmp-{uuid.uuid4().hex[:8]}"
            try:
                shutil.copytree(source_dir, temp_target, symlinks=False)
                temp_target.rename(target_dir)
            finally:
                if temp_target.exists():
                    shutil.rmtree(temp_target, ignore_errors=True)

    return thread_skills_root


def get_builtin_skill_specs() -> list[Any]:
    from yuxi.agents.skills.buildin import BUILTIN_SKILLS

    return BUILTIN_SKILLS


def _get_builtin_skill_spec_or_raise(slug: str) -> Any:
    normalized_slug = slug.strip() if isinstance(slug, str) else ""
    for spec in get_builtin_skill_specs():
        if getattr(spec, "slug", "").strip() == normalized_slug:
            return spec
    raise ValueError(f"内置 skill '{slug}' 不存在")


def _build_builtin_skill_dir_path(slug: str) -> str:
    return (Path("skills") / slug).as_posix()


def _is_builtin_managed(item: Skill, slug: str) -> bool:
    expected_dir = _build_builtin_skill_dir_path(slug)
    if item.dir_path != expected_dir:
        return False
    return (item.created_by or "") in {"system", BUILTIN_SKILL_OPERATOR}


def _dirs_equal(dir1: Path, dir2: Path) -> bool:
    """检查两个目录内容是否相同（通过文件列表比较）"""
    if not dir1.exists() or not dir2.exists():
        return False
    list1 = sorted([f.relative_to(dir1) for f in dir1.rglob("*") if f.is_file()])
    list2 = sorted([f.relative_to(dir2) for f in dir2.rglob("*") if f.is_file()])
    return list1 == list2


def _compute_dir_hash(source_dir: Path) -> str:
    hasher = hashlib.sha256()
    file_paths = sorted(path for path in source_dir.rglob("*") if path.is_file())
    for file_path in file_paths:
        relative_path = file_path.relative_to(source_dir).as_posix()
        hasher.update(relative_path.encode("utf-8"))
        hasher.update(b"\0")
        with file_path.open("rb") as f:
            while chunk := f.read(1024 * 1024):
                hasher.update(chunk)
        hasher.update(b"\0")
    return hasher.hexdigest()


def _copy_skill_target(target_dir: Path, source_dir: Path) -> None:
    if target_dir.is_symlink():
        target_dir.unlink()
    elif target_dir.exists():
        if _dirs_equal(target_dir, source_dir):
            return
        raise ValueError(f"技能目录已存在且非内置链接，无法托管: {target_dir}")

    shutil.copytree(source_dir, target_dir, symlinks=False, dirs_exist_ok=True)


def _replace_skill_target(target_dir: Path, source_dir: Path) -> None:
    temp_target = target_dir.with_name(f".{target_dir.name}.tmp-{uuid.uuid4().hex[:8]}")
    trash_dir: Path | None = None
    if temp_target.exists():
        shutil.rmtree(temp_target, ignore_errors=True)

    shutil.copytree(source_dir, temp_target, symlinks=False)
    try:
        if target_dir.exists():
            trash_dir = target_dir.with_name(f".{target_dir.name}.bak-{uuid.uuid4().hex[:8]}")
            target_dir.rename(trash_dir)
        temp_target.rename(target_dir)
    except Exception:
        shutil.rmtree(temp_target, ignore_errors=True)
        if trash_dir and trash_dir.exists() and not target_dir.exists():
            trash_dir.rename(target_dir)
        raise

    if trash_dir and trash_dir.exists():
        shutil.rmtree(trash_dir, ignore_errors=True)


async def get_skill_dependency_options(db: AsyncSession) -> dict[str, list[str] | list[dict]]:
    # 并行执行三个独立操作
    from yuxi.services.tool_service import get_tool_metadata

    async def get_skills():
        repo = SkillRepository(db)
        return await repo.list_all()

    def get_tools():
        all_tools = get_tool_metadata()
        return [{"id": tool["id"], "name": tool.get("name", tool["id"])} for tool in all_tools]

    items, tool_list, mcp_names = await asyncio.gather(
        get_skills(),
        asyncio.to_thread(get_tools),
        get_enabled_mcp_server_names(db=db),
    )

    return {
        "tools": tool_list,
        "mcps": mcp_names,
        "skills": [item.slug for item in items],
    }


async def list_skills(db: AsyncSession) -> list[Skill]:
    repo = SkillRepository(db)
    return await repo.list_all()


def _get_all_tool_names() -> list[str]:
    """获取所有工具名称（包括 buildin 和其他来源）"""
    from yuxi.services.tool_service import get_tool_metadata

    all_tools = get_tool_metadata()
    return [tool["id"] for tool in all_tools]


async def _validate_dependencies(
    *,
    slug: str,
    tool_dependencies: list[str],
    mcp_dependencies: list[str],
    skill_dependencies: list[str],
    available_skill_slugs: set[str],
) -> tuple[list[str], list[str], list[str]]:
    tools = _normalize_string_list(tool_dependencies)
    mcps = _normalize_string_list(mcp_dependencies)
    skills = _normalize_string_list(skill_dependencies)

    # 验证所有工具（不仅仅是 buildin）
    available_tools = set(_get_all_tool_names())
    invalid_tools = [name for name in tools if name not in available_tools]
    if invalid_tools:
        raise ValueError(f"存在无效工具依赖: {', '.join(invalid_tools)}")

    available_mcps = set(await get_enabled_mcp_server_names(db=None))
    invalid_mcps = [name for name in mcps if name not in available_mcps]
    if invalid_mcps:
        raise ValueError(f"存在无效 MCP 依赖: {', '.join(invalid_mcps)}")

    invalid_skills = [name for name in skills if name not in available_skill_slugs]
    if invalid_skills:
        raise ValueError(f"存在无效 skill 依赖: {', '.join(invalid_skills)}")

    if slug in skills:
        raise ValueError("skill_dependencies 不允许包含自身")

    return tools, mcps, skills


async def update_skill_dependencies(
    db: AsyncSession,
    *,
    slug: str,
    tool_dependencies: list[str],
    mcp_dependencies: list[str],
    skill_dependencies: list[str],
    updated_by: str | None,
) -> Skill:
    item = await get_skill_or_raise(db, slug)
    repo = SkillRepository(db)
    skill_items = await repo.list_all()
    available_skill_slugs = {skill.slug for skill in skill_items}
    tools, mcps, skills = await _validate_dependencies(
        slug=slug,
        tool_dependencies=tool_dependencies,
        mcp_dependencies=mcp_dependencies,
        skill_dependencies=skill_dependencies,
        available_skill_slugs=available_skill_slugs,
    )

    return await repo.update_dependencies(
        item,
        tool_dependencies=tools,
        mcp_dependencies=mcps,
        skill_dependencies=skills,
        updated_by=updated_by,
    )


def _validate_skill_name(name: str) -> str:
    name = name.strip()
    if not name:
        raise ValueError("SKILL.md frontmatter 缺少 name")
    if len(name) > 128:
        raise ValueError("skill name 长度不能超过 128")
    if not SKILL_NAME_PATTERN.match(name):
        raise ValueError("skill name 必须是小写字母/数字/短横线，且不能连续短横线")
    return name


def _split_frontmatter(content: str) -> tuple[str, str]:
    if not content.startswith("---"):
        raise ValueError("SKILL.md 缺少有效 frontmatter（--- ... ---）")

    lines = content.splitlines(keepends=True)
    if not lines or lines[0].strip() != "---":
        raise ValueError("SKILL.md 缺少有效 frontmatter（--- ... ---）")

    frontmatter_lines: list[str] = []
    body_start = 0
    for index, line in enumerate(lines[1:], start=1):
        if line.strip() == "---":
            body_start = index + 1
            break
        frontmatter_lines.append(line)
    else:
        raise ValueError("SKILL.md 缺少有效 frontmatter（--- ... ---）")

    frontmatter_raw = "".join(frontmatter_lines)
    body = "".join(lines[body_start:])
    return frontmatter_raw, body


def _parse_skill_markdown(content: str) -> tuple[str, str, dict[str, Any]]:
    frontmatter_raw, _body = _split_frontmatter(content)
    try:
        data = yaml.safe_load(frontmatter_raw)
    except yaml.YAMLError as e:
        raise ValueError(f"SKILL.md frontmatter YAML 解析失败: {e}") from e

    if not isinstance(data, dict):
        raise ValueError("SKILL.md frontmatter 必须是对象")

    name = _validate_skill_name(str(data.get("name", "")))
    description = str(data.get("description", "")).strip()
    if not description:
        raise ValueError("SKILL.md frontmatter 缺少 description")

    return name, description, data


def _rewrite_frontmatter_name(content: str, new_name: str) -> str:
    frontmatter_raw, body = _split_frontmatter(content)
    data = yaml.safe_load(frontmatter_raw)
    if not isinstance(data, dict):
        raise ValueError("SKILL.md frontmatter 必须是对象")
    data["name"] = new_name
    dumped = yaml.safe_dump(data, sort_keys=False, allow_unicode=True).strip()
    return f"---\n{dumped}\n---\n{body}"


def _validate_zip_paths(zip_file: zipfile.ZipFile) -> None:
    for name in zip_file.namelist():
        pure = PurePosixPath(name)
        if pure.is_absolute():
            raise ValueError(f"ZIP 包含不安全绝对路径: {name}")
        if ".." in pure.parts:
            raise ValueError(f"ZIP 包含路径穿越片段: {name}")


async def _generate_available_slug(repo: SkillRepository, base_slug: str) -> str:
    root = get_skills_root_dir()
    if not await repo.exists_slug(base_slug) and not (root / base_slug).exists():
        return base_slug

    idx = 2
    while True:
        candidate = f"{base_slug}-v{idx}"
        if not await repo.exists_slug(candidate) and not (root / candidate).exists():
            return candidate
        idx += 1


async def _import_skill_dir_impl(
    db: AsyncSession,
    *,
    source_skill_dir: Path,
    created_by: str | None,
) -> Skill:
    repo = SkillRepository(db)
    skills_root = get_skills_root_dir()

    skill_md_path = source_skill_dir / "SKILL.md"
    if not skill_md_path.exists() or not skill_md_path.is_file():
        raise ValueError("技能目录缺少根级 SKILL.md")

    content = skill_md_path.read_text(encoding="utf-8")
    parsed_name, parsed_desc, _ = _parse_skill_markdown(content)

    final_slug = await _generate_available_slug(repo, parsed_name)
    final_name = parsed_name
    with tempfile.TemporaryDirectory(prefix=".skill-import-", dir=str(skills_root.parent)) as temp_root:
        temp_root_path = Path(temp_root)
        stage_dir = temp_root_path / "stage"
        shutil.copytree(source_skill_dir, stage_dir)

        if final_slug != parsed_name:
            final_name = final_slug
            content = _rewrite_frontmatter_name(content, final_name)
            (stage_dir / "SKILL.md").write_text(content, encoding="utf-8")

        temp_target = skills_root / f".{final_slug}.tmp-{uuid.uuid4().hex[:8]}"
        if temp_target.exists():
            shutil.rmtree(temp_target)
        shutil.move(str(stage_dir), str(temp_target))

        final_dir = skills_root / final_slug
        if final_dir.exists():
            shutil.rmtree(temp_target, ignore_errors=True)
            raise ValueError(f"技能目录冲突，请重试: {final_slug}")
        temp_target.rename(final_dir)

        try:
            item = await repo.create(
                slug=final_slug,
                name=final_name,
                description=parsed_desc,
                tool_dependencies=[],
                mcp_dependencies=[],
                skill_dependencies=[],
                dir_path=(Path("skills") / final_slug).as_posix(),
                created_by=created_by,
            )
        except Exception:
            shutil.rmtree(final_dir, ignore_errors=True)
            raise

    return item


def _resolve_skill_dir(item: Skill) -> Path:
    dir_path = Path(item.dir_path)
    if dir_path.is_absolute():
        return dir_path
    return (Path(sys_config.save_dir) / dir_path).resolve()


def _resolve_relative_path(skill_dir: Path, relative_path: str, *, allow_root: bool = False) -> tuple[Path, str]:
    rel = (relative_path or "").strip().replace("\\", "/")
    rel = rel.lstrip("/")
    if not rel and not allow_root:
        raise ValueError("path 不能为空")
    pure = PurePosixPath(rel) if rel else PurePosixPath(".")
    if ".." in pure.parts:
        raise ValueError("非法路径：不允许上级路径引用")

    target = (skill_dir / pure).resolve()
    try:
        target.relative_to(skill_dir)
    except ValueError:
        raise ValueError("非法路径：越界访问被拒绝") from None

    return target, rel


def _is_text_path(path: Path) -> bool:
    if path.name == "SKILL.md":
        return True
    suffix = path.suffix.lower()
    return suffix in TEXT_FILE_EXTENSIONS


def _build_tree(path: Path, base_dir: Path) -> list[dict[str, Any]]:
    children: list[dict[str, Any]] = []
    for child in sorted(path.iterdir(), key=lambda p: (not p.is_dir(), p.name.lower())):
        rel = child.relative_to(base_dir).as_posix()
        if child.is_dir():
            children.append(
                {
                    "name": child.name,
                    "path": rel,
                    "is_dir": True,
                    "children": _build_tree(child, base_dir),
                }
            )
        else:
            children.append(
                {
                    "name": child.name,
                    "path": rel,
                    "is_dir": False,
                }
            )
    return children


async def import_skill_zip(
    db: AsyncSession,
    *,
    filename: str,
    file_bytes: bytes,
    created_by: str | None,
) -> Skill:
    normalized_filename = filename.lower()
    is_zip_upload = normalized_filename.endswith(".zip")
    is_skill_md_upload = normalized_filename.endswith("skill.md")
    if not is_zip_upload and not is_skill_md_upload:
        raise ValueError("仅支持上传 .zip 或 SKILL.md 文件")

    skills_root = get_skills_root_dir()

    with tempfile.TemporaryDirectory(prefix=".skill-import-", dir=str(skills_root.parent)) as temp_root:
        temp_root_path = Path(temp_root)
        extract_dir = temp_root_path / "extract"
        extract_dir.mkdir(parents=True, exist_ok=True)
        if is_zip_upload:
            zip_path = temp_root_path / "upload.zip"
            zip_path.write_bytes(file_bytes)

            with zipfile.ZipFile(zip_path, "r") as zf:
                _validate_zip_paths(zf)
                zf.extractall(extract_dir)

            skill_md_files = list(extract_dir.rglob("SKILL.md"))
            if len(skill_md_files) != 1:
                raise ValueError("ZIP 必须且只能包含一个技能（检测到一个 SKILL.md）")

            skill_md_path = skill_md_files[0]
            source_skill_dir = skill_md_path.parent
        else:
            source_skill_dir = extract_dir
            skill_md_path = source_skill_dir / "SKILL.md"
            skill_md_path.write_bytes(file_bytes)

        return await _import_skill_dir_impl(
            db,
            source_skill_dir=source_skill_dir,
            created_by=created_by,
        )


async def import_skill_dir(
    db: AsyncSession,
    *,
    source_dir: Path | str,
    created_by: str | None,
) -> Skill:
    source_skill_dir = Path(source_dir).resolve()
    # Confine to the system temp directory to prevent path traversal
    tmp_root = Path(tempfile.gettempdir()).resolve()
    if not source_skill_dir.is_relative_to(tmp_root):
        raise ValueError("技能目录路径不合法")
    if not source_skill_dir.exists() or not source_skill_dir.is_dir():
        raise ValueError("技能目录不存在")
    return await _import_skill_dir_impl(
        db,
        source_skill_dir=source_skill_dir,
        created_by=created_by,
    )


async def get_skill_or_raise(db: AsyncSession, slug: str) -> Skill:
    slug = slug.strip() if isinstance(slug, str) else ""
    if not is_valid_skill_slug(slug):
        raise ValueError("无效 skill slug")

    repo = SkillRepository(db)
    item = await repo.get_by_slug(slug)
    if not item:
        raise ValueError(f"技能 '{slug}' 不存在")
    return item


async def get_skill_tree(db: AsyncSession, slug: str) -> list[dict[str, Any]]:
    item = await get_skill_or_raise(db, slug)
    skill_dir = _resolve_skill_dir(item)
    if not skill_dir.exists() or not skill_dir.is_dir():
        raise ValueError(f"技能目录不存在: {item.dir_path}")
    return _build_tree(skill_dir, skill_dir)


async def read_skill_file(db: AsyncSession, slug: str, relative_path: str) -> dict[str, Any]:
    item = await get_skill_or_raise(db, slug)
    skill_dir = _resolve_skill_dir(item)
    target, rel = _resolve_relative_path(skill_dir, relative_path)
    if not target.exists() or not target.is_file():
        raise ValueError(f"文件不存在: {relative_path}")
    if not _is_text_path(target):
        raise ValueError("仅支持读取文本文件")
    try:
        content = target.read_text(encoding="utf-8")
    except UnicodeDecodeError as e:
        raise ValueError(f"文件编码不支持（仅支持 UTF-8）: {e}") from e

    return {"path": rel, "content": content}


async def create_skill_node(
    db: AsyncSession,
    *,
    slug: str,
    relative_path: str,
    is_dir: bool,
    content: str | None,
    updated_by: str | None,
) -> None:
    item = await get_skill_or_raise(db, slug)
    if item.is_builtin:
        raise ValueError("内置 skill 不允许直接修改文件")
    skill_dir = _resolve_skill_dir(item)
    target, _ = _resolve_relative_path(skill_dir, relative_path)
    if target.exists():
        raise ValueError("目标已存在")

    if is_dir:
        target.mkdir(parents=True, exist_ok=False)
        return

    if not _is_text_path(target):
        raise ValueError("仅支持创建文本文件")

    target.parent.mkdir(parents=True, exist_ok=True)

    # 先写入文件，再更新元数据
    target.write_text(content or "", encoding="utf-8")

    await _update_skill_metadata_if_skills_md(db, item, content or "", skill_dir, target, updated_by)


async def update_skill_file(
    db: AsyncSession,
    *,
    slug: str,
    relative_path: str,
    content: str,
    updated_by: str | None,
) -> None:
    item = await get_skill_or_raise(db, slug)
    if item.is_builtin:
        raise ValueError("内置 skill 不允许直接修改文件")
    skill_dir = _resolve_skill_dir(item)
    target, _ = _resolve_relative_path(skill_dir, relative_path)
    if not target.exists() or not target.is_file():
        raise ValueError("文件不存在")
    if not _is_text_path(target):
        raise ValueError("仅支持编辑文本文件")

    await _update_skill_metadata_if_skills_md(db, item, content, skill_dir, target, updated_by)

    target.write_text(content, encoding="utf-8")


async def _update_skill_metadata_if_skills_md(
    db: AsyncSession,
    item: Skill,
    content: str,
    skill_dir: Path,
    target: Path,
    updated_by: str | None,
) -> None:
    """如果目标文件是 SKILL.md，则解析并更新元数据"""
    if target.name == "SKILL.md" and target.parent == skill_dir:
        parsed_name, parsed_desc, _ = _parse_skill_markdown(content)
        if parsed_name != item.slug:
            raise ValueError("SKILL.md frontmatter.name 必须与 skill slug 一致")
        repo = SkillRepository(db)
        await repo.update_metadata(item, name=parsed_name, description=parsed_desc, updated_by=updated_by)


async def delete_skill_node(db: AsyncSession, *, slug: str, relative_path: str) -> None:
    item = await get_skill_or_raise(db, slug)
    if item.is_builtin:
        raise ValueError("内置 skill 不允许直接修改文件")
    skill_dir = _resolve_skill_dir(item)
    target, rel = _resolve_relative_path(skill_dir, relative_path, allow_root=False)
    if not target.exists():
        raise ValueError("目标不存在")

    if rel == "SKILL.md":
        raise ValueError("不允许删除根目录 SKILL.md")

    if target.is_dir():
        shutil.rmtree(target)
    else:
        target.unlink()


async def export_skill_zip(db: AsyncSession, slug: str) -> tuple[str, str]:
    item = await get_skill_or_raise(db, slug)
    skill_dir = _resolve_skill_dir(item)
    if not skill_dir.exists() or not skill_dir.is_dir():
        raise ValueError("技能目录不存在")

    fd, export_path = tempfile.mkstemp(prefix=f"skill-{slug}-", suffix=".zip")
    Path(export_path).unlink(missing_ok=True)
    export_file = Path(export_path)
    try:
        with zipfile.ZipFile(export_file, "w", compression=zipfile.ZIP_DEFLATED) as zf:
            for p in skill_dir.rglob("*"):
                arcname = Path(slug) / p.relative_to(skill_dir)
                zf.write(p, arcname.as_posix())
    except Exception:
        export_file.unlink(missing_ok=True)
        raise
    return export_path, f"{slug}.zip"


async def delete_skill(db: AsyncSession, *, slug: str) -> None:
    repo = SkillRepository(db)
    item = await repo.get_by_slug(slug)
    if not item:
        raise ValueError(f"技能 '{slug}' 不存在")

    skill_dir = _resolve_skill_dir(item)
    trash_dir: Path | None = None

    if skill_dir.exists():
        trash_dir = skill_dir.with_name(f".deleted-{slug}-{uuid.uuid4().hex[:8]}")
        skill_dir.rename(trash_dir)

    try:
        await repo.delete(item)
    except Exception:
        if trash_dir and trash_dir.exists():
            trash_dir.rename(skill_dir)
        raise

    if trash_dir and trash_dir.exists():
        shutil.rmtree(trash_dir, ignore_errors=True)


async def init_builtin_skills(db: AsyncSession, *, created_by: str = "system") -> None:
    """校验内置 skills 配置，不执行安装。"""
    specs = get_builtin_skill_specs()

    for spec in specs:
        slug = str(getattr(spec, "slug", "")).strip()
        source_dir = Path(str(getattr(spec, "source_dir", ""))).resolve()

        if not is_valid_skill_slug(slug):
            raise ValueError(f"内置 skill slug 非法: {slug}")
        if not source_dir.exists() or not source_dir.is_dir():
            logger.warning(f"跳过不存在的内置 skill 目录: {source_dir}")
            continue

        skill_md = source_dir / "SKILL.md"
        if not skill_md.exists():
            legacy_skill_md = source_dir / "SKILLS.md"
            if not legacy_skill_md.exists():
                raise ValueError(f"内置 skill 缺少 SKILL.md: {source_dir}")
            skill_md = legacy_skill_md

        content = skill_md.read_text(encoding="utf-8")
        parsed_name, _, meta = _parse_skill_markdown(content)
        if parsed_name != slug:
            raise ValueError(f"内置 skill frontmatter.name 必须等于 slug: {slug}")
        _normalize_string_list(meta.get("tool_dependencies"))
        _normalize_string_list(meta.get("mcp_dependencies"))
        _normalize_string_list(meta.get("skill_dependencies"))
        _compute_dir_hash(source_dir)


def list_builtin_skill_specs() -> list[dict[str, Any]]:
    specs: list[dict[str, Any]] = []
    for raw_spec in get_builtin_skill_specs():
        slug = str(getattr(raw_spec, "slug", "")).strip()
        source_dir = Path(str(getattr(raw_spec, "source_dir", ""))).resolve()
        configured_description = str(getattr(raw_spec, "description", "")).strip()
        version = str(getattr(raw_spec, "version", "1.0.0")).strip() or "1.0.0"
        configured_tools = _normalize_string_list(getattr(raw_spec, "tool_dependencies", None))
        configured_mcps = _normalize_string_list(getattr(raw_spec, "mcp_dependencies", None))
        configured_skills = _normalize_string_list(getattr(raw_spec, "skill_dependencies", None))

        if not is_valid_skill_slug(slug):
            raise ValueError(f"内置 skill slug 非法: {slug}")
        if not source_dir.exists() or not source_dir.is_dir():
            logger.warning(f"跳过不存在的内置 skill 目录: {source_dir}")
            continue

        skill_md = source_dir / "SKILL.md"
        if not skill_md.exists():
            legacy_skill_md = source_dir / "SKILLS.md"
            if not legacy_skill_md.exists():
                raise ValueError(f"内置 skill 缺少 SKILL.md: {source_dir}")
            skill_md = legacy_skill_md

        content = skill_md.read_text(encoding="utf-8")
        parsed_name, parsed_desc, meta = _parse_skill_markdown(content)
        if parsed_name != slug:
            raise ValueError(f"内置 skill frontmatter.name 必须等于 slug: {slug}")

        specs.append(
            {
                "slug": slug,
                "name": slug,
                "description": configured_description or parsed_desc,
                "version": version,
                "tool_dependencies": configured_tools or _normalize_string_list(meta.get("tool_dependencies")),
                "mcp_dependencies": configured_mcps or _normalize_string_list(meta.get("mcp_dependencies")),
                "skill_dependencies": configured_skills or _normalize_string_list(meta.get("skill_dependencies")),
                "content_hash": _compute_dir_hash(source_dir),
                "source_dir": source_dir,
            }
        )

    return specs


async def install_builtin_skill(db: AsyncSession, slug: str, *, installed_by: str | None) -> Skill:
    _get_builtin_skill_spec_or_raise(slug)
    repo = SkillRepository(db)
    spec = next(item for item in list_builtin_skill_specs() if item["slug"] == slug)

    existing = await repo.get_by_slug(slug)
    if existing:
        raise ValueError(f"内置 skill '{slug}' 已安装")

    target_dir = get_skills_root_dir() / slug
    if target_dir.exists():
        raise ValueError(f"技能目录已存在: {slug}")

    shutil.copytree(Path(spec["source_dir"]), target_dir, symlinks=False)
    try:
        return await repo.create(
            slug=slug,
            name=spec["name"],
            description=spec["description"],
            tool_dependencies=spec["tool_dependencies"],
            mcp_dependencies=spec["mcp_dependencies"],
            skill_dependencies=spec["skill_dependencies"],
            dir_path=_build_builtin_skill_dir_path(slug),
            version=spec["version"],
            is_builtin=True,
            content_hash=spec["content_hash"],
            created_by=installed_by or BUILTIN_SKILL_OPERATOR,
        )
    except Exception:
        shutil.rmtree(target_dir, ignore_errors=True)
        raise


async def update_builtin_skill(
    db: AsyncSession,
    slug: str,
    *,
    force: bool = False,
    updated_by: str | None,
) -> Skill:
    _get_builtin_skill_spec_or_raise(slug)
    repo = SkillRepository(db)
    spec = next(item for item in list_builtin_skill_specs() if item["slug"] == slug)
    item = await repo.get_by_slug(slug)
    if not item:
        raise ValueError(f"内置 skill '{slug}' 未安装")
    if not item.is_builtin and not _is_builtin_managed(item, slug):
        raise ValueError(f"技能 '{slug}' 不是内置 skill")

    if item.content_hash != spec["content_hash"] and not force:
        raise BuiltinSkillUpdateConflictError("检测到你修改过此 skill，更新将覆盖你的修改，是否继续？")

    target_dir = _resolve_skill_dir(item)
    _replace_skill_target(target_dir, Path(spec["source_dir"]))

    if item.name != spec["name"] or item.description != spec["description"]:
        await repo.update_metadata(
            item,
            name=spec["name"],
            description=spec["description"],
            updated_by=updated_by,
        )

    if (
        _normalize_string_list(item.tool_dependencies or []) != spec["tool_dependencies"]
        or _normalize_string_list(item.mcp_dependencies or []) != spec["mcp_dependencies"]
        or _normalize_string_list(item.skill_dependencies or []) != spec["skill_dependencies"]
    ):
        await repo.update_dependencies(
            item,
            tool_dependencies=spec["tool_dependencies"],
            mcp_dependencies=spec["mcp_dependencies"],
            skill_dependencies=spec["skill_dependencies"],
            updated_by=updated_by,
        )

    return await repo.update_builtin_install(
        item,
        version=spec["version"],
        content_hash=spec["content_hash"],
        updated_by=updated_by,
    )