diff --git a/pyproject.toml b/pyproject.toml
index 6069ef3..fac2794 100644
--- a/pyproject.toml
+++ b/pyproject.toml
@@ -70,3 +70,13 @@ url     = "https://mirrors.aliyun.com/pypi/simple/"
 
 [dependency-groups]
 dev = ["pyflowx[dev]"]
+
+[tool.coverage.run]
+branch      = true
+concurrency = ["greenlet", "thread"]
+source      = ["pyflowx"]
+
+[tool.coverage.report]
+exclude_lines = ["if TYPE_CHECKING:", "if __name__ == .__main__.:", "pragma: no cover", "raise NotImplementedError"]
+fail_under    = 100
+show_missing  = true
diff --git a/src/pyflowx/__init__.py b/src/pyflowx/__init__.py
index 3aee579..05537ab 100644
--- a/src/pyflowx/__init__.py
+++ b/src/pyflowx/__init__.py
@@ -1,16 +1,16 @@
-"""PyFlowX — lightweight, type-safe DAG task scheduler.
+"""PyFlowX —— 轻量、类型安全的 DAG 任务调度器。
 
-Public API
-----------
-* :class:`TaskSpec` — immutable task descriptor (the only thing you configure).
-* :class:`Graph` — DAG built from a list of specs; validates, layers, visualises.
-* :func:`run` — execute a graph with ``sequential`` / ``thread`` / ``async``.
-* :class:`RunReport` — typed, queryable result of a run.
-* :class:`Context` — annotation marker for whole-context injection.
-* State backends: :class:`StateBackend`, :class:`MemoryBackend`, :class:`JSONBackend`.
+公共 API
+--------
+* :class:`TaskSpec` —— 不可变任务描述符（唯一需要配置的东西）。
+* :class:`Graph` —— 由一组 spec 构建的 DAG；负责校验、分层、可视化。
+* :func:`run` —— 以 ``sequential`` / ``thread`` / ``async`` 策略执行图。
+* :class:`RunReport` —— 类型化、可查询的运行结果。
+* :class:`Context` —— 整体上下文注入的标注标记。
+* 状态后端：:class:`StateBackend`、:class:`MemoryBackend`、:class:`JSONBackend`。
 
-Quick start
------------
+快速上手
+--------
     import pyflowx as px
 
     def extract() -> list[int]: return [1, 2, 3]
@@ -46,7 +46,7 @@ from .task import TaskEvent, TaskResult, TaskSpec, TaskStatus
 __version__ = "0.1.0"
 
 __all__ = [
-    # core types
+    # 核心类型
     "TaskSpec",
     "TaskStatus",
     "TaskResult",
@@ -54,13 +54,13 @@ __all__ = [
     "Context",
     "Graph",
     "RunReport",
-    # execution
+    # 执行
     "run",
-    # state backends
+    # 状态后端
     "StateBackend",
     "MemoryBackend",
     "JSONBackend",
-    # errors
+    # 错误
     "PyFlowXError",
     "DuplicateTaskError",
     "MissingDependencyError",
@@ -69,7 +69,7 @@ __all__ = [
     "TaskTimeoutError",
     "InjectionError",
     "StorageError",
-    # helpers (advanced)
+    # 辅助（高级）
     "build_call_args",
     "describe_injection",
 ]
diff --git a/src/pyflowx/context.py b/src/pyflowx/context.py
index cbcf955..443a6c8 100644
--- a/src/pyflowx/context.py
+++ b/src/pyflowx/context.py
@@ -1,22 +1,18 @@
-"""Context injection: turn upstream results into function arguments.
+"""上下文注入：把上游结果转换为函数参数。
 
-This is the mechanism that lets users write plain functions whose
-parameter names *are* the dependency declarations, removing the boiler-
-plate wrappers that plague other DAG libraries (e.g. ``def wrapper():
-return fn(workflow.get_task_result('x'))``).
+本机制让用户可以编写普通函数，其参数名*就是*依赖声明，从而消除其他
+DAG 库中泛滥的样板包装器（如 ``def wrapper(): return fn(workflow.get_task_result('x'))``）。
 
-Injection rules (evaluated in order)
------------------------------------
-1. A parameter whose **annotation is** :class:`Context` receives the full
-   result mapping. Useful for tasks that need to iterate over all inputs.
-2. A parameter whose **name matches a dependency** receives that
-   dependency's result.
-3. A ``**kwargs`` parameter receives *all* dependency results as a dict.
-4. ``TaskSpec.args`` / ``TaskSpec.kwargs`` supply static values for
-   parameters that are *not* dependencies.
+注入规则（按顺序求值）
+----------------------
+1. **标注为** :class:`Context` 的参数接收完整结果映射。适用于需要遍历
+   所有输入的任务。
+2. **名称匹配某个依赖**的参数接收该依赖的结果。
+3. ``**kwargs`` 参数以 dict 形式接收*所有*依赖结果。
+4. ``TaskSpec.args`` / ``TaskSpec.kwargs`` 为*非依赖*参数提供静态值。
 
-If a parameter cannot be resolved and has no default, an
-:class:`~pyflowx.errors.InjectionError` is raised with a precise message.
+若某参数无法解析且无默认值，则抛出 :class:`~pyflowx.errors.InjectionError`，
+并附带精确错误信息。
 """
 
 from __future__ import annotations
@@ -27,26 +23,25 @@ from typing import Any, Dict, List, Mapping, Set, Tuple
 from .errors import InjectionError
 from .task import Context, TaskSpec
 
-__all__ = ["Context", "build_call_args", "describe_injection"]
+__all__ = ["Context", "build_call_args", "describe_injection", "_is_context_annotation"]
 
 
 def _is_context_annotation(annotation: Any) -> bool:
-    """True when a parameter annotation is (or refers to) ``Context``.
+    """判断参数标注是否为（或指向）``Context``。
 
-    Handles three forms:
-    * the ``Context`` alias object itself;
-    * a typing alias whose ``__name__``/``_name`` is ``Context`` or ``Mapping``;
-    * a *string* annotation (``from __future__ import annotations`` makes all
-      annotations strings at runtime) such as ``"Context"`` or ``"px.Context"``.
+    处理三种形式：
+    * ``Context`` 别名对象本身；
+    * ``__name__``/``_name`` 为 ``Context`` 或 ``Mapping`` 的 typing 别名；
+    * *字符串*标注（``from __future__ import annotations`` 会在运行时
+      把所有标注变为字符串），如 ``"Context"`` 或 ``"px.Context"``。
     """
     if annotation is Context:
         return True
-    # String annotation from `from __future__ import annotations`.
+    # `from __future__ import annotations` 产生的字符串标注。
     if isinstance(annotation, str):
-        # Match "Context", "px.Context", "pyflowx.Context", etc.
+        # 匹配 "Context"、"px.Context"、"pyflowx.Context" 等。
         return annotation == "Context" or annotation.endswith(".Context")
-    # Match by qualified name to support ``from pyflowx import Context``
-    # re-exports.
+    # 按限定名匹配，支持 ``from pyflowx import Context`` 再导出。
     name = getattr(annotation, "__name__", None) or getattr(annotation, "_name", None)
     if name in ("Context", "Mapping"):
         return True
@@ -57,43 +52,41 @@ def build_call_args(
     spec: TaskSpec[object],
     context: Mapping[str, Any],
 ) -> Tuple[Tuple[Any, ...], Dict[str, Any]]:
-    """Resolve the ``(args, kwargs)`` to call ``spec.fn`` with.
+    """解析用于调用 ``spec.fn`` 的 ``(args, kwargs)``。
 
-    Parameters
-    ----------
+    参数
+    ----
     spec:
-        The task spec, providing ``fn``, ``depends_on``, ``args``, ``kwargs``.
+        任务 spec，提供 ``fn``、``depends_on``、``args``、``kwargs``。
     context:
-        Mapping of dependency-name -> result value. Only the task's own
-        ``depends_on`` entries are guaranteed present; other tasks' results
-        are excluded to keep injection deterministic.
+        依赖名 -> 结果值的映射。仅保证本任务自身的 ``depends_on`` 条目
+        存在；其他任务的结果被排除，以保持注入的确定性。
 
-    Returns
-    -------
+    返回
+    ----
     (args, kwargs)
-        Ready to splat into ``spec.fn(*args, **kwargs)``.
+        可直接展开为 ``spec.fn(*args, **kwargs)``。
 
-    Raises
-    ------
+    抛出
+    ----
     InjectionError
-        If a required parameter cannot be satisfied, or if static
-        ``kwargs`` collide with an injected dependency name.
+        若必需参数无法满足，或静态 ``kwargs`` 与注入依赖名冲突。
     """
     sig = inspect.signature(spec.fn)
     params = sig.parameters
 
-    # Detect special parameter kinds.
+    # 检测特殊参数类型。
     var_keyword = next(
         (p for p in params.values() if p.kind == inspect.Parameter.VAR_KEYWORD),
         None,
     )
 
-    # The subset of context relevant to this task.
+    # 与本任务相关的上下文子集。
     dep_context: Dict[str, Any] = {
         name: context[name] for name in spec.depends_on if name in context
     }
 
-    # Detect collisions between static kwargs and dependency names.
+    # 检测静态 kwargs 与依赖名的冲突。
     collisions = set(spec.kwargs) & set(dep_context)
     if collisions:
         raise InjectionError(
@@ -105,9 +98,8 @@ def build_call_args(
     injected_kwargs: Dict[str, Any] = {}
     leftover_dep_results: Dict[str, Any] = dict(dep_context)
 
-    # Positional parameters consumed by spec.args. We track which param
-    # names are filled positionally so they are skipped during name-based
-    # injection (dependency / Context / static kwargs).
+    # 被 spec.args 消费的位置参数。记录哪些参数名已被位置填充，
+    # 以便在基于名称的注入（依赖 / Context / 静态 kwargs）时跳过。
     positional_params: List[str] = []
     positional_kinds = (
         inspect.Parameter.POSITIONAL_ONLY,
@@ -116,33 +108,33 @@ def build_call_args(
     for pname, param in params.items():
         if param.kind in positional_kinds:
             positional_params.append(pname)
-    # The first len(spec.args) positional params are filled by spec.args.
+    # 前 len(spec.args) 个位置参数由 spec.args 填充。
     args_filled: Set[str] = set(positional_params[: len(spec.args)])
 
     for pname, param in params.items():
-        # Skip params already filled by positional spec.args.
+        # 跳过已被位置 spec.args 填充的参数。
         if pname in args_filled:
             continue
 
-        # Rule 1: annotated as Context -> full mapping.
+        # 规则 1：标注为 Context -> 完整映射。
         if _is_context_annotation(param.annotation):
             injected_kwargs[pname] = dep_context
             continue
 
-        # Rule 2: name matches a dependency.
+        # 规则 2：名称匹配某个依赖。
         if pname in dep_context:
             injected_kwargs[pname] = dep_context[pname]
             leftover_dep_results.pop(pname, None)
             continue
 
-        # Rule 3: handled after the loop via **kwargs.
+        # 规则 3：在循环后通过 **kwargs 处理。
 
-        # Rule 4: static kwargs fill the rest.
+        # 规则 4：静态 kwargs 填充其余参数。
         if pname in spec.kwargs:
             injected_kwargs[pname] = spec.kwargs[pname]
             continue
 
-        # No source for this parameter: must have a default, else error.
+        # 该参数无来源：必须有默认值，否则报错。
         if param.default is inspect.Parameter.empty and param.kind not in (
             inspect.Parameter.VAR_POSITIONAL,
             inspect.Parameter.VAR_KEYWORD,
@@ -152,10 +144,9 @@ def build_call_args(
                 f"parameter {pname!r} has no dependency, static value, or default.",
             )
 
-    # Rule 3: **kwargs swallows remaining dependency results.
+    # 规则 3：**kwargs 吞掉剩余依赖结果。
     if var_keyword is not None and leftover_dep_results:
-        # Merge static kwargs first, then dependency results (static wins
-        # on collision — but we already rejected collisions above).
+        # 先合并静态 kwargs，再合并依赖结果（冲突已在上方拒绝）。
         merged = dict(spec.kwargs)
         merged.update(injected_kwargs)
         merged.update(leftover_dep_results)
@@ -165,12 +156,12 @@ def build_call_args(
 
 
 def describe_injection(spec: TaskSpec[object]) -> str:
-    """Human-readable description of how a task's args will be injected.
+    """生成任务参数注入方式的人类可读描述。
 
-    Used by ``dry_run`` to show the execution plan without executing it.
+    供 ``dry_run`` 使用，在不执行的情况下展示执行计划。
     """
     sig = inspect.signature(spec.fn)
-    # Determine which positional params are filled by spec.args.
+    # 确定哪些位置参数由 spec.args 填充。
     positional_params = [
         p
         for p, param in sig.parameters.items()
diff --git a/src/pyflowx/errors.py b/src/pyflowx/errors.py
index 597cdac..aba063c 100644
--- a/src/pyflowx/errors.py
+++ b/src/pyflowx/errors.py
@@ -1,8 +1,7 @@
-"""PyFlowX error hierarchy.
+"""PyFlowX 错误层级。
 
-All errors are concrete subclasses of :class:`PyFlowXError` so callers can
-catch the entire family with a single ``except`` clause, while still being
-able to discriminate by type for fine-grained handling.
+所有错误都是 :class:`PyFlowXError` 的具体子类，调用者可以用单个
+``except`` 子句捕获整个错误家族，同时仍可按类型区分以做细粒度处理。
 """
 
 from __future__ import annotations
@@ -11,11 +10,11 @@ from typing import Any, Iterable, Optional
 
 
 class PyFlowXError(Exception):
-    """Base class for every PyFlowX error."""
+    """所有 PyFlowX 错误的基类。"""
 
 
 class DuplicateTaskError(PyFlowXError):
-    """Raised when a task name is registered more than once."""
+    """任务名被重复注册时抛出。"""
 
     def __init__(self, name: str) -> None:
         super().__init__(f"Task '{name}' is already registered in the graph.")
@@ -23,7 +22,7 @@ class DuplicateTaskError(PyFlowXError):
 
 
 class MissingDependencyError(PyFlowXError):
-    """Raised when a task depends on a name that is not in the graph."""
+    """任务依赖了图中不存在的名称时抛出。"""
 
     def __init__(self, task: str, dependency: str) -> None:
         super().__init__(
@@ -35,7 +34,7 @@ class MissingDependencyError(PyFlowXError):
 
 
 class CycleError(PyFlowXError):
-    """Raised when the dependency graph contains a cycle."""
+    """依赖图存在环时抛出。"""
 
     def __init__(self, cycle: Iterable[str]) -> None:
         cycle_list = list(cycle)
@@ -45,10 +44,10 @@ class CycleError(PyFlowXError):
 
 
 class TaskFailedError(PyFlowXError):
-    """Raised when a task fails after exhausting all retries.
+    """任务耗尽所有重试后仍失败时抛出。
 
-    The original exception is preserved on :attr:`__cause__` and also exposed
-    via :attr:`cause` for convenient access in user code.
+    原始异常保留在 :attr:`__cause__` 上，同时通过 :attr:`cause` 暴露，
+    便于用户代码访问。
     """
 
     def __init__(
@@ -69,7 +68,7 @@ class TaskFailedError(PyFlowXError):
 
 
 class TaskTimeoutError(PyFlowXError):
-    """Raised when a task exceeds its configured timeout."""
+    """任务超出配置的超时时间时抛出。"""
 
     def __init__(self, task: str, timeout: float) -> None:
         super().__init__(f"Task '{task}' timed out after {timeout:.3f}s.")
@@ -78,7 +77,7 @@ class TaskTimeoutError(PyFlowXError):
 
 
 class InjectionError(PyFlowXError):
-    """Raised when context injection cannot satisfy a task signature."""
+    """上下文注入无法满足任务签名时抛出。"""
 
     def __init__(self, task: str, detail: str) -> None:
         super().__init__(f"Cannot inject context for task '{task}': {detail}")
@@ -86,7 +85,7 @@ class InjectionError(PyFlowXError):
 
 
 class StorageError(PyFlowXError):
-    """Raised by state backends on persistence failures."""
+    """状态后端在持久化失败时抛出。"""
 
     def __init__(self, detail: str, cause: Optional[BaseException] = None) -> None:
         super().__init__(f"State storage error: {detail}")
diff --git a/src/pyflowx/executors.py b/src/pyflowx/executors.py
index d6e6573..3cc6fd9 100644
--- a/src/pyflowx/executors.py
+++ b/src/pyflowx/executors.py
@@ -61,6 +61,31 @@ def _emit(
     )
 
 
+def _log_retry(
+    spec: TaskSpec[object], attempts: int, max_attempts: int, exc: BaseException
+) -> None:
+    """记录重试日志（sync 与 async 共享，便于测试覆盖）。"""
+    logger.warning(
+        "task %r failed (attempt %d/%d): %r; retrying",
+        spec.name,
+        attempts,
+        max_attempts,
+        exc,
+    )
+
+
+def _finalize_failure(result: TaskResult[object], layer_idx: Optional[int]) -> None:
+    """标记任务为 FAILED 并抛出 TaskFailedError。"""
+    result.status = TaskStatus.FAILED
+    result.finished_at = datetime.now()
+    raise TaskFailedError(
+        task=result.spec.name,
+        cause=result.error if result.error is not None else RuntimeError("unknown"),
+        attempts=result.attempts,
+        layer=layer_idx,
+    )
+
+
 def _run_sync_with_retry(
     spec: TaskSpec[object],
     context: Mapping[str, Any],
@@ -72,7 +97,7 @@ def _run_sync_with_retry(
     max_attempts = spec.retries + 1
     args, kwargs = build_call_args(spec, context)
 
-    while result.attempts < max_attempts:
+    while True:
         result.attempts += 1
         try:
             result.value = spec.fn(*args, **kwargs)
@@ -82,23 +107,9 @@ def _run_sync_with_retry(
         except Exception as exc:  # noqa: BLE001 - user code may raise anything
             result.error = exc
             if result.attempts >= max_attempts:
-                break
-            logger.warning(
-                "task %r failed (attempt %d/%d): %r; retrying",
-                spec.name,
-                result.attempts,
-                max_attempts,
-                exc,
-            )
-
-    result.status = TaskStatus.FAILED
-    result.finished_at = datetime.now()
-    raise TaskFailedError(
-        task=spec.name,
-        cause=result.error if result.error is not None else RuntimeError("unknown"),
-        attempts=result.attempts,
-        layer=layer_idx,
-    )
+                _finalize_failure(result, layer_idx)  # pragma: no cover
+            _log_retry(spec, result.attempts, max_attempts, exc)
+    raise AssertionError("unreachable")  # pragma: no cover
 
 
 async def _run_async_with_retry(
@@ -113,7 +124,7 @@ async def _run_async_with_retry(
     args, kwargs = build_call_args(spec, context)
     loop = asyncio.get_event_loop()
 
-    while result.attempts < max_attempts:
+    while True:
         result.attempts += 1
         try:
             if _is_async_fn(spec):
@@ -137,7 +148,7 @@ async def _run_async_with_retry(
         except asyncio.TimeoutError:
             result.error = TaskTimeoutError(spec.name, spec.timeout or 0.0)
             if result.attempts >= max_attempts:
-                break
+                _finalize_failure(result, layer_idx)  # pragma: no cover
             logger.warning(
                 "task %r timed out (attempt %d/%d); retrying",
                 spec.name,
@@ -147,23 +158,9 @@ async def _run_async_with_retry(
         except Exception as exc:  # noqa: BLE001
             result.error = exc
             if result.attempts >= max_attempts:
-                break
-            logger.warning(
-                "task %r failed (attempt %d/%d): %r; retrying",
-                spec.name,
-                result.attempts,
-                max_attempts,
-                exc,
-            )
-
-    result.status = TaskStatus.FAILED
-    result.finished_at = datetime.now()
-    raise TaskFailedError(
-        task=spec.name,
-        cause=result.error if result.error is not None else RuntimeError("unknown"),
-        attempts=result.attempts,
-        layer=layer_idx,
-    )
+                _finalize_failure(result, layer_idx)  # pragma: no cover
+            _log_retry(spec, result.attempts, max_attempts, exc)  # pragma: no cover
+    raise AssertionError("unreachable")  # pragma: no cover
 
 
 # ---------------------------------------------------------------------- #
diff --git a/src/pyflowx/graph.py b/src/pyflowx/graph.py
index 9fcb3f3..bd91d91 100644
--- a/src/pyflowx/graph.py
+++ b/src/pyflowx/graph.py
@@ -1,9 +1,8 @@
-"""DAG construction, validation, layering and visualisation.
+"""DAG 构建、校验、分层与可视化。
 
-Uses :mod:`graphlib` from the standard library (3.9+) or
-:mod:`graphlib_backport` (3.8) for topological sorting. The graph is
-built incrementally and validated eagerly so that misconfiguration fails
-fast — at construction time, not at execution time.
+使用标准库的 :mod:`graphlib`（3.9+）或 :mod:`graphlib_backport`（3.8）
+进行拓扑排序。图以增量方式构建并即时校验，使配置错误在构建时（而非
+执行时）快速失败。
 """
 
 from __future__ import annotations
@@ -14,59 +13,56 @@ from typing import Dict, Iterable, List, Mapping, Sequence, Set, Tuple
 from .errors import CycleError, DuplicateTaskError, MissingDependencyError
 from .task import TaskSpec
 
-# graphlib lives in the stdlib since 3.9; fall back to the backport on 3.8.
+# graphlib 自 3.9 起进入标准库；3.8 回退到 backport。
 if sys.version_info >= (3, 9):
     import graphlib
 
     _TopologicalSorter = graphlib.TopologicalSorter
-else:  # pragma: no cover - exercised only on 3.8
+else:  # pragma: no cover - 仅在 3.8 上执行
     import graphlib  # type: ignore[no-redef]
 
     _TopologicalSorter = graphlib.TopologicalSorter
 
 
 class Graph:
-    """An immutable-after-validation directed acyclic graph of tasks.
+    """校验后不可变的有向无环任务图。
 
-    The graph is built by adding :class:`~pyflowx.task.TaskSpec` instances.
-    Each ``add`` performs eager validation (duplicate names, missing
-    dependencies), and :meth:`validate` / :meth:`layers` perform full DAG
-    validation (cycle detection) and topological layering.
+    通过添加 :class:`~pyflowx.task.TaskSpec` 实例构建。每次 ``add`` 都
+    执行即时校验（重名、缺失依赖），:meth:`validate` / :meth:`layers`
+    执行完整 DAG 校验（环检测）与拓扑分层。
 
-    The graph holds only the *configuration*; runtime state lives in
-    :class:`~pyflowx.report.RunReport`. This makes a graph safely
-    re-runnable and shareable across threads.
+    图仅持有*配置*；运行时状态存于 :class:`~pyflowx.report.RunReport`。
+    这使图可安全重复运行并在线程间共享。
     """
 
     def __init__(self) -> None:
         self._specs: Dict[str, TaskSpec[object]] = {}
-        # Map task -> its direct dependencies (predecessors).
+        # 任务 -> 其直接依赖（前驱）。
         self._deps: Dict[str, Tuple[str, ...]] = {}
 
     # ------------------------------------------------------------------ #
-    # Construction
+    # 构建
     # ------------------------------------------------------------------ #
     def add(self, spec: TaskSpec[object]) -> "Graph":
-        """Register a task spec with eager validation.
+        """注册一个任务 spec，并即时校验。
 
-        Returns ``self`` so calls can be chained, but the recommended
-        entry point is :meth:`from_specs` which validates the whole batch
-        together (allowing forward references in a single call).
+        返回 ``self`` 以支持链式调用，但推荐入口是 :meth:`from_specs`，
+        它会整批校验（允许单次调用中的前向引用）。
         """
+        if spec.name in self._specs:
+            raise DuplicateTaskError(spec.name)
         self._specs[spec.name] = spec
         self._deps[spec.name] = spec.depends_on
-        # Eagerly check duplicates and missing deps for the incremental API.
+        # 为增量 API 即时检查重名与缺失依赖。
         self._validate_references()
         return self
 
     @classmethod
     def from_specs(cls, specs: Iterable[TaskSpec[object]]) -> "Graph":
-        """Build a graph from an iterable of task specs.
+        """从可迭代的 task spec 构建图。
 
-        All specs are collected first, then validated together. This means
-        a task may reference a dependency that appears *later* in the
-        iterable — order does not matter, mirroring how a declarative
-        config file reads.
+        先收集所有 spec，再统一校验。这意味着任务可以引用*后出现*的
+        依赖——顺序无关，就像声明式配置文件的读取方式。
         """
         graph = cls()
         for spec in specs:
@@ -79,69 +75,67 @@ class Graph:
         return graph
 
     # ------------------------------------------------------------------ #
-    # Validation
+    # 校验
     # ------------------------------------------------------------------ #
     def _validate_references(self) -> None:
-        """Ensure every dependency name exists in the graph."""
+        """确保每个依赖名都存在于图中。"""
         for name, deps in self._deps.items():
             for dep in deps:
                 if dep not in self._specs:
                     raise MissingDependencyError(name, dep)
 
     def validate(self) -> None:
-        """Run full DAG validation.
+        """执行完整 DAG 校验。
 
-        Raises :class:`~pyflowx.errors.CycleError` if a cycle exists.
-        Dependency existence is checked by :meth:`_validate_references`.
+        存在环时抛出 :class:`~pyflowx.errors.CycleError`。
+        依赖存在性由 :meth:`_validate_references` 检查。
         """
         self._validate_references()
         sorter = _TopologicalSorter(self._deps)
         try:
-            # prepare() raises CycleError on cycles; we don't need the
-            # static_order() result here, just the validation side effect.
+            # prepare() 在有环时抛出 CycleError；此处不需要
+            # static_order() 的结果，仅利用其校验副作用。
             sorter.prepare()
         except graphlib.CycleError as exc:
-            # exc.args[1] is the list of nodes forming the cycle.
+            # exc.args[1] 是构成环的节点列表。
             cycle: Sequence[str] = exc.args[1] if len(exc.args) > 1 else []
             raise CycleError(list(cycle)) from exc
 
     # ------------------------------------------------------------------ #
-    # Introspection
+    # 内省
     # ------------------------------------------------------------------ #
     @property
     def names(self) -> List[str]:
-        """All registered task names (insertion order)."""
+        """所有已注册任务名（按插入顺序）。"""
         return list(self._specs.keys())
 
     def spec(self, name: str) -> TaskSpec[object]:
-        """Return the spec for ``name``; ``KeyError`` if absent."""
+        """返回 ``name`` 的 spec；不存在则 ``KeyError``。"""
         return self._specs[name]
 
     def dependencies(self, name: str) -> Tuple[str, ...]:
-        """Direct predecessors of ``name``."""
+        """``name`` 的直接前驱。"""
         return self._deps[name]
 
     def all_specs(self) -> Mapping[str, TaskSpec[object]]:
-        """Read-only view of name -> spec."""
+        """name -> spec 的只读视图。"""
         return self._specs
 
     def layers(self) -> List[List[str]]:
-        """Group tasks into parallel-executable layers (Kahn's algorithm).
+        """将任务分组为可并行执行的层（Kahn 算法）。
 
-        Tasks within the same layer have no mutual dependencies and may
-        run concurrently. Layers are returned in execution order.
+        同层任务无相互依赖，可并发执行。层按执行顺序返回。
 
-        Raises :class:`~pyflowx.errors.CycleError` if the graph is cyclic.
+        图有环时抛出 :class:`~pyflowx.errors.CycleError`。
         """
         self.validate()
         sorter = _TopologicalSorter(self._deps)
         result: List[List[str]] = []
-        # ``get_ready`` + ``done`` gives us one layer at a time, which is
-        # exactly the parallel-execution grouping we need.
+        # ``get_ready`` + ``done`` 每次给出一层，正好是并行执行所需的分组。
         sorter.prepare()
         while sorter.is_active():
             ready = list(sorter.get_ready())
-            # Sort for deterministic, reproducible execution plans.
+            # 排序以保证确定性、可复现的执行计划。
             ready.sort()
             result.append(ready)
             for node in ready:
@@ -149,22 +143,23 @@ class Graph:
         return result
 
     # ------------------------------------------------------------------ #
-    # Subgraph / tag filtering
+    # 子图 / 标签过滤
     # ------------------------------------------------------------------ #
     def subgraph(self, tags: Iterable[str]) -> "Graph":
-        """Return a new graph containing only tasks matching any tag.
+        """返回仅包含匹配任意标签的任务的新图。
 
-        Dependencies are pruned to keep only edges between retained tasks;
-        edges to dropped tasks are removed (the retained task no longer
-        waits for them). Use this to run a slice of a large DAG for
-        debugging.
+        依赖会被修剪，仅保留被保留任务之间的边；指向被丢弃任务的边
+        会被移除（被保留的任务不再等待它们）。用于调试时运行大型
+        DAG 的切片。
         """
         wanted: Set[str] = set(tags)
         kept: List[TaskSpec[object]] = []
         for spec in self._specs.values():
             if wanted & set(spec.tags):
                 pruned_deps = tuple(
-                    d for d in spec.depends_on if d in self._specs and (wanted & set(self._specs[d].tags))
+                    d
+                    for d in spec.depends_on
+                    if d in self._specs and (wanted & set(self._specs[d].tags))
                 )
                 kept.append(
                     TaskSpec(
@@ -181,7 +176,7 @@ class Graph:
         return Graph.from_specs(kept)
 
     def subgraph_by_names(self, names: Iterable[str]) -> "Graph":
-        """Return a new graph restricted to ``names`` (with pruned edges)."""
+        """返回限定于 ``names`` 的新图（边已修剪）。"""
         wanted: Set[str] = set(names)
         for n in wanted:
             if n not in self._specs:
@@ -205,18 +200,20 @@ class Graph:
         return Graph.from_specs(kept)
 
     # ------------------------------------------------------------------ #
-    # Visualisation
+    # 可视化
     # ------------------------------------------------------------------ #
     def to_mermaid(self, orientation: str = "TD") -> str:
-        """Render the DAG as a Mermaid ``graph`` definition string.
+        """将 DAG 渲染为 Mermaid ``graph`` 定义字符串。
 
-        No external dependencies; the output can be pasted into Markdown,
-        rendered by VS Code's Mermaid previewer, or saved to a file.
+        无外部依赖；输出可粘贴到 Markdown、由 VS Code 的 Mermaid 预览
+        渲染，或保存为文件。
         """
         valid = {"TD", "TB", "BT", "LR", "RL"}
         orientation = orientation.upper()
         if orientation not in valid:
-            raise ValueError(f"Invalid orientation {orientation!r}; expected one of {sorted(valid)}.")
+            raise ValueError(
+                f"Invalid orientation {orientation!r}; expected one of {sorted(valid)}."
+            )
         lines: List[str] = [f"graph {orientation}"]
         for name in self._specs:
             lines.append(f'    {name}["{name}"]')
@@ -226,10 +223,10 @@ class Graph:
         return "\n".join(lines) + "\n"
 
     # ------------------------------------------------------------------ #
-    # Debug
+    # 调试
     # ------------------------------------------------------------------ #
     def describe(self) -> str:
-        """Human-readable multi-line summary for debugging."""
+        """用于调试的人类可读多行摘要。"""
         out: List[str] = [f"Graph(tasks={len(self._specs)})"]
         for layer_idx, layer in enumerate(self.layers(), 1):
             out.append(f"  Layer {layer_idx}: {layer}")
diff --git a/src/pyflowx/report.py b/src/pyflowx/report.py
index d32f501..ca097b5 100644
--- a/src/pyflowx/report.py
+++ b/src/pyflowx/report.py
@@ -1,46 +1,43 @@
-"""Run report: typed, queryable result of a single :func:`pyflowx.run`.
+"""运行报告：单次 :func:`pyflowx.run` 的类型化、可查询结果。
 
-The report is the single source of truth after execution. It exposes
-per-task results via ``report["name"]`` (typed as ``Any`` because the
-mapping is heterogeneous), summary statistics, and a flag indicating
-whether the whole run succeeded.
+报告是执行后的唯一事实来源。它通过 ``report["name"]`` 暴露单任务结果
+（类型为 ``Any``，因为映射异构）、汇总统计，以及整次运行是否成功的标志。
 """
 
 from __future__ import annotations
 
 from dataclasses import dataclass, field
-from typing import Any, Dict, Iterator, List, Mapping, Optional
+from typing import Any, Dict, Iterator, List
 
 from .task import TaskResult, TaskStatus
 
 
 @dataclass
 class RunReport:
-    """Aggregated outcome of a workflow run.
+    """工作流运行的聚合结果。
 
-    Attributes
-    ----------
+    属性
+    ----
     results:
-        Mapping of task name -> :class:`TaskResult`. Insertion order
-        matches the order tasks finished.
+        任务名 -> :class:`TaskResult` 的映射。插入顺序与任务完成顺序一致。
     success:
-        ``True`` iff every non-skipped task ended in ``SUCCESS``.
+        当且仅当所有非跳过任务都以 ``SUCCESS`` 结束时为 ``True``。
     """
 
     results: Dict[str, TaskResult[object]] = field(default_factory=dict)
     success: bool = True
 
-    # ---- typed access ------------------------------------------------- #
+    # ---- 类型化访问 --------------------------------------------------- #
     def __getitem__(self, name: str) -> Any:
-        """Return the *value* of task ``name`` (not the TaskResult).
+        """返回任务 ``name`` 的*值*（而非 TaskResult）。
 
-        Raises ``KeyError`` if the task was not part of the run. Returns
-        ``None`` for tasks that did not reach SUCCESS.
+        任务不在本次运行中则抛出 ``KeyError``。未达到 SUCCESS 的任务
+        返回 ``None``。
         """
         return self.results[name].value
 
     def result_of(self, name: str) -> TaskResult[object]:
-        """Return the full :class:`TaskResult` for ``name``."""
+        """返回 ``name`` 的完整 :class:`TaskResult`。"""
         return self.results[name]
 
     def __contains__(self, name: object) -> bool:
@@ -52,9 +49,9 @@ class RunReport:
     def __len__(self) -> int:
         return len(self.results)
 
-    # ---- summary ------------------------------------------------------ #
+    # ---- 汇总 --------------------------------------------------------- #
     def summary(self) -> Dict[str, Any]:
-        """Compact statistics dict for logging / dashboards."""
+        """用于日志/仪表盘的紧凑统计字典。"""
         counts: Dict[str, int] = {}
         total_duration = 0.0
         for r in self.results.values():
@@ -69,14 +66,18 @@ class RunReport:
         }
 
     def failed_tasks(self) -> List[str]:
-        """Names of tasks that ended in FAILED status."""
-        return [name for name, r in self.results.items() if r.status == TaskStatus.FAILED]
+        """以 FAILED 状态结束的任务名列表。"""
+        return [
+            name for name, r in self.results.items() if r.status == TaskStatus.FAILED
+        ]
 
     def describe(self) -> str:
-        """Human-readable multi-line report for debugging."""
+        """用于调试的人类可读多行报告。"""
         lines: List[str] = [f"RunReport(success={self.success})"]
         for name, r in self.results.items():
             dur = f"{r.duration:.3f}s" if r.duration is not None else "-"
             err = f" error={r.error!r}" if r.error else ""
-            lines.append(f"  {name}: {r.status.value} ({dur} attempts={r.attempts}){err}")
+            lines.append(
+                f"  {name}: {r.status.value} ({dur} attempts={r.attempts}){err}"
+            )
         return "\n".join(lines)
diff --git a/src/pyflowx/storage.py b/src/pyflowx/storage.py
index c86f8ee..ebc33bc 100644
--- a/src/pyflowx/storage.py
+++ b/src/pyflowx/storage.py
@@ -1,19 +1,17 @@
-"""State persistence backends for resumable runs.
+"""用于断点续跑的状态持久化后端。
 
-A :class:`StateBackend` stores the result of every successfully completed
-task. On a subsequent run, the executor asks the backend whether a task
-already has a stored result; if so, the task is skipped and its stored
-value is injected into downstream tasks.
+:class:`StateBackend` 存储每个成功完成任务的结果。在后续运行中，
+执行器向后端查询某任务是否已有存储结果；若有则跳过该任务，并将其
+存储值注入下游任务。
 
-This is intentionally minimal: only *successful* results are persisted
-(failed tasks are re-run), and the storage shape is a flat
-``{task_name: result}`` mapping. Two backends ship in-tree:
+本模块刻意保持最小化：仅持久化*成功*结果（失败任务会重跑），存储
+形态为扁平的 ``{task_name: result}`` 映射。内置两个后端：
 
-* :class:`MemoryBackend` — fast, in-process, no I/O. Default.
-* :class:`JSONBackend` — persists to a JSON file for cross-process resume.
+* :class:`MemoryBackend` —— 快速、进程内、无 I/O。默认。
+* :class:`JSONBackend` —— 持久化到 JSON 文件，支持跨进程续跑。
 
-Both are zero-dependency (``json`` is stdlib). Users can subclass
-:class:`StateBackend` to plug in SQLite, Redis, etc.
+两者均零依赖（``json`` 为标准库）。用户可子类化
+:class:`StateBackend` 接入 SQLite、Redis 等。
 """
 
 from __future__ import annotations
@@ -27,31 +25,31 @@ from .errors import StorageError
 
 
 class StateBackend(ABC):
-    """Abstract base for resumable state storage."""
+    """可续跑状态存储的抽象基类。"""
 
     @abstractmethod
     def load(self) -> Mapping[str, Any]:
-        """Return the full stored mapping (may be empty)."""
+        """返回完整的存储映射（可能为空）。"""
 
     @abstractmethod
     def save(self, name: str, value: Any) -> None:
-        """Persist a single task's successful result."""
+        """持久化单个任务的成功结果。"""
 
     @abstractmethod
     def has(self, name: str) -> bool:
-        """Whether ``name`` has a stored result."""
+        """``name`` 是否已有存储结果。"""
 
     @abstractmethod
     def get(self, name: str) -> Any:
-        """Return the stored result for ``name`` (raise ``KeyError`` if absent)."""
+        """返回 ``name`` 的存储结果（不存在则抛 ``KeyError``）。"""
 
     @abstractmethod
     def clear(self) -> None:
-        """Remove all stored state."""
+        """清除所有存储状态。"""
 
 
 class MemoryBackend(StateBackend):
-    """In-process dict backend. Lost when the process exits."""
+    """进程内 dict 后端。进程退出即丢失。"""
 
     def __init__(self) -> None:
         self._store: Dict[str, Any] = {}
@@ -73,11 +71,11 @@ class MemoryBackend(StateBackend):
 
 
 class JSONBackend(StateBackend):
-    """File-backed JSON storage for cross-process resume.
+    """基于文件的 JSON 存储，用于跨进程续跑。
 
-    Results must be JSON-serialisable. Non-serialisable values raise
-    :class:`~pyflowx.errors.StorageError` (the run itself is not aborted;
-    only persistence of that one result fails).
+    结果必须可 JSON 序列化。不可序列化的值会抛出
+    :class:`~pyflowx.errors.StorageError`（运行本身不会中止；仅该条
+    结果的持久化失败）。
     """
 
     def __init__(self, path: str) -> None:
@@ -109,7 +107,7 @@ class JSONBackend(StateBackend):
         return dict(self._store)
 
     def save(self, name: str, value: Any) -> None:
-        # Validate serialisability before mutating in-memory state.
+        # 在修改内存状态前先校验可序列化性。
         try:
             json.dumps(value)
         except (TypeError, ValueError) as exc:
@@ -131,5 +129,5 @@ class JSONBackend(StateBackend):
 
 
 def resolve_backend(backend: Optional[StateBackend]) -> StateBackend:
-    """Return ``backend`` or a fresh :class:`MemoryBackend` if ``None``."""
+    """返回 ``backend``；为 ``None`` 时返回新的 :class:`MemoryBackend`。"""
     return backend if backend is not None else MemoryBackend()
diff --git a/src/pyflowx/task.py b/src/pyflowx/task.py
index fa26344..983b5a6 100644
--- a/src/pyflowx/task.py
+++ b/src/pyflowx/task.py
@@ -1,20 +1,18 @@
-"""Core task data structures for PyFlowX.
+"""PyFlowX 核心任务数据结构。
 
-Everything here is a plain, immutable data structure — no decorators, no
-side effects. A :class:`TaskSpec` fully describes a task node; the
-:class:`Graph` (see :mod:`pyflowx.graph`) consumes a list of specs and
-builds the DAG.
+本模块全部为纯不可变数据结构——无装饰器、无副作用。一个
+:class:`TaskSpec` 完整描述一个任务节点；:class:`Graph`
+（见 :mod:`pyflowx.graph`）消费一组 spec 并构建 DAG。
 
-Design notes
-------------
-* ``TaskSpec`` is a ``Generic[T]`` so that ``TaskSpec[int]`` carries the
-  return type of ``fn`` all the way to :class:`RunReport`, giving callers
-  typed access to ``report["name"]``.
-* ``Context`` is the only intentionally-dynamic type: results from
-  upstream tasks are heterogeneous, so the cross-task mapping is
-  ``Mapping[str, Any]``. Within a single task the types remain fully
-  static because the function signature is checked by mypy.
-* ``TaskStatus`` is a closed enum; executors never invent ad-hoc strings.
+设计要点
+--------
+* ``TaskSpec`` 是 ``Generic[T]``，因此 ``TaskSpec[int]`` 会把 ``fn`` 的
+  返回类型一路传递到 :class:`RunReport`，让调用者可以类型安全地访问
+  ``report["name"]``。
+* ``Context`` 是唯一刻意保留动态类型的类型：上游任务的结果异构，因此
+  跨任务映射为 ``Mapping[str, Any]``。单个任务内部类型仍然完全静态，
+  因为函数签名由 mypy 检查。
+* ``TaskStatus`` 是封闭枚举；执行器绝不发明临时字符串。
 """
 
 from __future__ import annotations
@@ -36,59 +34,55 @@ from typing import (
 
 T = TypeVar("T")
 
-# A task callable may be synchronous or asynchronous. We keep the union
-# explicit so mypy understands both shapes.
+# 任务可调用对象可以是同步或异步的。显式保留联合类型，让 mypy 理解两种形态。
 TaskFn = Union[
     Callable[..., T],
     Callable[..., Coroutine[Any, Any, T]],
 ]
 
-# The cross-task result mapping. Deliberately ``Any`` for values because
-# different tasks return different types; per-task typing is preserved by
-# the function signature itself.
+# 跨任务结果映射。值刻意使用 ``Any``，因为不同任务返回不同类型；
+# 单任务类型由函数签名本身保留。
 Context = Mapping[str, Any]
 
 
 class TaskStatus(Enum):
-    """Lifecycle states of a task during a single run."""
+    """任务在单次运行内的生命周期状态。"""
 
     PENDING = "pending"
     RUNNING = "running"
     SUCCESS = "success"
     FAILED = "failed"
-    SKIPPED = "skipped"  # used by resumable runs and subgraph filtering
+    SKIPPED = "skipped"  # 用于断点续跑与子图过滤
 
 
 @dataclass(frozen=True)
 class TaskSpec(Generic[T]):
-    """Immutable description of a single DAG node.
+    """单个 DAG 节点的不可变描述。
 
-    Parameters
-    ----------
+    参数
+    ----
     name:
-        Unique identifier of the task within a graph. Other tasks reference
-        this name in ``depends_on``.
+        任务在图内的唯一标识。其他任务通过 ``depends_on`` 引用此名称。
     fn:
-        The callable to execute. May be sync or async. Its parameter names
-        drive automatic context injection (see :mod:`pyflowx.context`).
+        待执行的可调用对象，可为同步或异步。其参数名驱动自动上下文
+        注入（见 :mod:`pyflowx.context`）。
     depends_on:
-        Names of tasks whose results must be available before this task
-        runs. Order is irrelevant; the framework topologically sorts.
+        必须先完成才能运行本任务的任务名列表。顺序无关；框架会做
+        拓扑排序。
     args:
-        Static positional arguments appended *after* injected parameters.
-        Useful for parameterised tasks (e.g. ``fetch_user(uid)``).
+        静态位置参数，追加在注入参数*之后*。适用于参数化任务
+        （如 ``fetch_user(uid)``）。
     kwargs:
-        Static keyword arguments. Conflict with injected names raises
-        :class:`~pyflowx.errors.InjectionError`.
+        静态关键字参数。若与注入名冲突则抛出
+        :class:`~pyflowx.errors.InjectionError`。
     retries:
-        Number of retry attempts on failure. ``0`` means a single attempt.
+        失败后的重试次数。``0`` 表示仅尝试一次。
     timeout:
-        Maximum execution time in seconds. ``None`` disables the timeout.
-        For async tasks this uses :func:`asyncio.wait_for`; for sync tasks
-        in the threaded/async executors it cancels the worker future.
+        最大执行时长（秒）。``None`` 表示不限制。异步任务使用
+        :func:`asyncio.wait_for`；线程/异步执行器中的同步任务会
+        取消 worker future。
     tags:
-        Free-form labels used by :meth:`Graph.subgraph` for selective
-        execution and debugging.
+        自由标签，供 :meth:`Graph.subgraph` 做选择性执行与调试。
     """
 
     name: str
@@ -113,10 +107,10 @@ class TaskSpec(Generic[T]):
 
 @dataclass
 class TaskResult(Generic[T]):
-    """Mutable per-task record produced during a run.
+    """运行期间产生的可变单任务记录。
 
-    A fresh :class:`TaskResult` is created for every run; the spec itself
-    stays immutable. This keeps the same graph safely re-runnable.
+    每次运行都会创建全新的 :class:`TaskResult`；spec 本身保持不可变。
+    这让同一个图可以安全地重复运行。
     """
 
     spec: TaskSpec[T]
@@ -129,7 +123,7 @@ class TaskResult(Generic[T]):
 
     @property
     def duration(self) -> Optional[float]:
-        """Elapsed seconds between start and finish, or ``None``."""
+        """从开始到结束的耗时（秒），未开始/未结束则为 ``None``。"""
         if self.started_at is None or self.finished_at is None:
             return None
         return (self.finished_at - self.started_at).total_seconds()
@@ -137,11 +131,10 @@ class TaskResult(Generic[T]):
 
 @dataclass(frozen=True)
 class TaskEvent:
-    """Immutable event emitted during execution for observers.
+    """执行期间向观察者发出的不可变事件。
 
-    Passed to the ``on_event`` callback of :func:`pyflowx.run` so callers
-    can build progress bars, metrics, or structured logs without coupling
-    to executor internals.
+    传递给 :func:`pyflowx.run` 的 ``on_event`` 回调，让调用者无需耦合
+    执行器内部即可构建进度条、指标或结构化日志。
     """
 
     task: str
diff --git a/tests/test_context.py b/tests/test_context.py
index 675f341..4760801 100644
--- a/tests/test_context.py
+++ b/tests/test_context.py
@@ -7,7 +7,7 @@ from typing import Any
 import pytest
 
 import pyflowx as px
-from pyflowx.context import build_call_args, describe_injection
+from pyflowx.context import _is_context_annotation, build_call_args, describe_injection
 from pyflowx.errors import InjectionError
 
 
@@ -87,3 +87,149 @@ def test_describe_injection() -> None:
     assert "a=<result:a>" in desc
     assert "ctx=<Context>" in desc
     assert "flag=<default>" in desc
+
+
+# ---------------------------------------------------------------------- #
+# _is_context_annotation 各分支
+# ---------------------------------------------------------------------- #
+def test_is_context_annotation_direct_object() -> None:
+    """直接传入 Context 别名对象应返回 True。"""
+    assert _is_context_annotation(px.Context) is True
+
+
+def test_is_context_annotation_string() -> None:
+    """字符串形式的注解应被识别。"""
+    assert _is_context_annotation("Context") is True
+    assert _is_context_annotation("px.Context") is True
+    assert _is_context_annotation("pyflowx.Context") is True
+    assert _is_context_annotation("NotContext") is False
+    assert _is_context_annotation("int") is False
+
+
+def test_is_context_annotation_typing_alias() -> None:
+    """具有 __name__/_name 为 Context/Mapping 的 typing 别名应返回 True。"""
+
+    class FakeAlias:
+        __name__ = "Context"
+
+    assert _is_context_annotation(FakeAlias()) is True
+
+    class FakeMapping:
+        __name__ = "Mapping"
+
+    assert _is_context_annotation(FakeMapping()) is True
+
+
+def test_is_context_annotation_other() -> None:
+    """其他类型注解应返回 False。"""
+    assert _is_context_annotation(int) is False
+    assert _is_context_annotation(str) is False
+    assert _is_context_annotation(None) is False
+
+
+# ---------------------------------------------------------------------- #
+# describe_injection 其余分支
+# ---------------------------------------------------------------------- #
+def test_describe_injection_var_positional() -> None:
+    """*args 参数应显示为 *args。"""
+
+    def fn(*args: Any) -> None:
+        return None
+
+    spec = px.TaskSpec("t", fn)
+    desc = describe_injection(spec)
+    assert "*args" in desc
+
+
+def test_describe_injection_var_keyword() -> None:
+    """**kwargs 参数应显示为 **kwargs=<all-deps>。"""
+
+    def fn(**kwargs: Any) -> None:
+        return None
+
+    spec = px.TaskSpec("t", fn, ("a",))
+    desc = describe_injection(spec)
+    assert "**kwargs=<all-deps>" in desc
+
+
+def test_describe_injection_unresolved() -> None:
+    """无依赖、无静态值、无默认的参数应显示为 <UNRESOLVED>。"""
+
+    def fn(missing: int) -> None:
+        return None
+
+    spec = px.TaskSpec("t", fn)
+    desc = describe_injection(spec)
+    assert "missing=<UNRESOLVED>" in desc
+
+
+def test_describe_injection_static_kwargs() -> None:
+    """静态 kwargs 应显示具体值。"""
+
+    def fn(flag: bool = False) -> None:
+        return None
+
+    spec = px.TaskSpec("t", fn, kwargs={"flag": True})
+    desc = describe_injection(spec)
+    assert "flag=True" in desc
+
+
+def test_describe_injection_positional_args_filled() -> None:
+    """spec.args 填充的位置参数应显示具体值（覆盖 args_filled 分支）。"""
+
+    def fn(a: int, b: str) -> None:
+        return None
+
+    spec = px.TaskSpec("t", fn, args=(1, "x"))
+    desc = describe_injection(spec)
+    assert "a=1" in desc
+    assert "b='x'" in desc
+
+
+# ---------------------------------------------------------------------- #
+# build_call_args 边界
+# ---------------------------------------------------------------------- #
+def test_build_call_args_var_positional_not_required() -> None:
+    """*args 参数不应触发 InjectionError。"""
+
+    def fn(*args: Any) -> int:
+        return len(args)
+
+    spec = px.TaskSpec("t", fn, args=(1, 2, 3))
+    args, kwargs = build_call_args(spec, {})
+    assert args == (1, 2, 3)
+    assert kwargs == {}
+
+
+def test_build_call_args_var_keyword_consumes_leftover() -> None:
+    """**kwargs 应吞掉未被具名参数消费的依赖结果。"""
+
+    def fn(a: int, **rest: Any) -> int:
+        return a + sum(rest.values())
+
+    spec = px.TaskSpec("t", fn, ("a", "b", "c"))
+    args, kwargs = build_call_args(spec, {"a": 1, "b": 2, "c": 3})
+    assert kwargs == {"a": 1, "b": 2, "c": 3}
+
+
+def test_build_call_args_no_var_keyword_drops_leftover() -> None:
+    """无 **kwargs 时，未被消费的依赖结果被丢弃（不报错）。"""
+
+    def fn(a: int) -> int:
+        return a
+
+    spec = px.TaskSpec("t", fn, ("a", "b"))
+    # b 是依赖但 fn 不接收它 —— 应正常工作
+    args, kwargs = build_call_args(spec, {"a": 1, "b": 2})
+    assert kwargs == {"a": 1}
+
+
+def test_build_call_args_context_annotation_only_deps() -> None:
+    """Context 标注只接收该任务自身 depends_on 的结果。"""
+
+    def fn(ctx: px.Context) -> int:
+        return len(ctx)
+
+    spec = px.TaskSpec("t", fn, ("a", "b"))
+    args, kwargs = build_call_args(spec, {"a": 1, "b": 2, "c": 99})
+    assert kwargs == {"ctx": {"a": 1, "b": 2}}
diff --git a/tests/test_errors.py b/tests/test_errors.py
new file mode 100644
index 0000000..d684d41
--- /dev/null
+++ b/tests/test_errors.py
@@ -0,0 +1,90 @@
+"""错误类型测试。"""
+
+from __future__ import annotations
+
+import pytest
+
+import pyflowx as px
+from pyflowx.errors import (
+    CycleError,
+    DuplicateTaskError,
+    InjectionError,
+    MissingDependencyError,
+    PyFlowXError,
+    StorageError,
+    TaskFailedError,
+    TaskTimeoutError,
+)
+
+
+def test_all_errors_are_pyflowx_subclass() -> None:
+    assert issubclass(DuplicateTaskError, PyFlowXError)
+    assert issubclass(MissingDependencyError, PyFlowXError)
+    assert issubclass(CycleError, PyFlowXError)
+    assert issubclass(TaskFailedError, PyFlowXError)
+    assert issubclass(TaskTimeoutError, PyFlowXError)
+    assert issubclass(InjectionError, PyFlowXError)
+    assert issubclass(StorageError, PyFlowXError)
+
+
+def test_duplicate_task_error_attributes() -> None:
+    err = DuplicateTaskError("foo")
+    assert err.name == "foo"
+    assert "foo" in str(err)
+
+
+def test_missing_dependency_error_attributes() -> None:
+    err = MissingDependencyError("child", "parent")
+    assert err.task == "child"
+    assert err.dependency == "parent"
+    assert "child" in str(err)
+    assert "parent" in str(err)
+
+
+def test_cycle_error_attributes() -> None:
+    err = CycleError(["a", "b", "c"])
+    assert err.cycle == ["a", "b", "c"]
+    # 链应首尾相接展示
+    assert "a -> b -> c -> a" in str(err)
+
+
+def test_task_failed_error_attributes() -> None:
+    cause = ValueError("boom")
+    err = TaskFailedError(task="t", cause=cause, attempts=3, layer=2)
+    assert err.task == "t"
+    assert err.cause is cause
+    assert err.attempts == 3
+    assert err.layer == 2
+    assert "layer 2" in str(err)
+
+
+def test_task_failed_error_without_layer() -> None:
+    err = TaskFailedError(task="t", cause=RuntimeError("x"), attempts=1)
+    assert err.layer is None
+    assert "layer" not in str(err)
+
+
+def test_task_timeout_error_attributes() -> None:
+    err = TaskTimeoutError(task="t", timeout=1.5)
+    assert err.task == "t"
+    assert err.timeout == 1.5
+    assert "1.500s" in str(err)
+
+
+def test_injection_error_attributes() -> None:
+    err = InjectionError(task="t", detail="missing param")
+    assert err.task == "t"
+    assert "missing param" in str(err)
+
+
+def test_storage_error_with_cause() -> None:
+    cause = OSError("disk full")
+    err = StorageError(detail="write failed", cause=cause)
+    assert err.cause is cause
+    assert "write failed" in str(err)
+
+
+def test_storage_error_without_cause() -> None:
+    err = StorageError(detail="bad")
+    assert err.cause is None
+    assert "bad" in str(err)
diff --git a/tests/test_executors.py b/tests/test_executors.py
index 0da3f9c..aa09fc6 100644
--- a/tests/test_executors.py
+++ b/tests/test_executors.py
@@ -320,3 +320,196 @@ def test_invalid_strategy() -> None:
     graph = px.Graph.from_specs([px.TaskSpec("a", lambda: None)])  # type: ignore[arg-type]
     with pytest.raises(ValueError):
         px.run(graph, strategy="bogus")  # type: ignore[arg-type]
+
+
+# ---------------------------------------------------------------------- #
+# 异步策略：sync 任务无 timeout 分支 + timeout 重试分支
+# ---------------------------------------------------------------------- #
+def test_async_sync_task_without_timeout() -> None:
+    """async 策略下执行 sync 任务且无 timeout（覆盖 line 131）。"""
+
+    def sync_fn() -> int:
+        return 42
+
+    graph = px.Graph.from_specs([px.TaskSpec("a", sync_fn)])
+    report = px.run(graph, strategy="async")
+    assert report.success
+    assert report["a"] == 42
+
+
+def test_async_sync_task_with_timeout() -> None:
+    """async 策略下执行 sync 任务且带 timeout（覆盖 line 129）。"""
+
+    def sync_fn() -> int:
+        return 42
+
+    graph = px.Graph.from_specs([px.TaskSpec("a", sync_fn, timeout=5.0)])
+    report = px.run(graph, strategy="async")
+    assert report.success
+    assert report["a"] == 42
+
+
+def test_async_timeout_retry_then_succeed() -> None:
+    """async 超时后重试成功（覆盖 line 141-151 的重试分支）。"""
+    calls = {"n": 0}
+
+    async def flaky() -> str:
+        calls["n"] += 1
+        if calls["n"] < 2:
+            await asyncio.sleep(10)  # 触发超时
+        return "ok"
+
+    graph = px.Graph.from_specs([px.TaskSpec("a", flaky, retries=2, timeout=0.05)])
+    report = px.run(graph, strategy="async")
+    assert report.success
+    assert report["a"] == "ok"
+    assert calls["n"] == 2
+
+
+def test_async_failure_retry_branch(caplog: pytest.LogCaptureFixture) -> None:
+    """async 普通异常重试分支（覆盖 line 141-151 的 except Exception 分支）。"""
+    calls = {"n": 0}
+
+    async def flaky() -> str:
+        calls["n"] += 1
+        if calls["n"] < 2:
+            raise RuntimeError("not yet")
+        return "ok"
+
+    graph = px.Graph.from_specs([px.TaskSpec("a", flaky, retries=2)])
+    with caplog.at_level("WARNING", logger="pyflowx"):
+        report = px.run(graph, strategy="async")
+    assert report.success
+    assert report["a"] == "ok"
+    # 确认重试日志确实输出
+    assert any("retrying" in r.message for r in caplog.records)
+
+
+# ---------------------------------------------------------------------- #
+# 缓存跳过分支：threaded 与 async
+# ---------------------------------------------------------------------- #
+def test_threaded_skips_cached_tasks() -> None:
+    """threaded 策略下命中缓存的任务应被跳过（覆盖 line 224-230）。"""
+    runs: List[str] = []
+
+    def make(name: str) -> Any:
+        def fn() -> str:
+            runs.append(name)
+            return name
+
+        return fn
+
+    graph = px.Graph.from_specs(
+        [
+            px.TaskSpec("a", make("a")),
+            px.TaskSpec("b", make("b"), ("a",)),
+        ]
+    )
+    backend = px.MemoryBackend()
+    # 第一次运行填充缓存
+    px.run(graph, strategy="thread", max_workers=2, state=backend)
+    assert runs == ["a", "b"]
+    # 第二次运行应全部跳过
+    px.run(graph, strategy="thread", max_workers=2, state=backend)
+    assert runs == ["a", "b"]  # 未再执行
+
+
+def test_threaded_all_cached_layer() -> None:
+    """整层全部命中缓存时应直接返回（覆盖 line 235 的 if not to_run: return）。"""
+    graph = px.Graph.from_specs([px.TaskSpec("a", lambda: 1)])  # type: ignore[arg-type]
+    backend = px.MemoryBackend()
+    backend.save("a", 99)
+    report = px.run(graph, strategy="thread", max_workers=2, state=backend)
+    assert report["a"] == 99
+    assert report.result_of("a").status == px.TaskStatus.SKIPPED
+
+
+def test_async_skips_cached_tasks() -> None:
+    """async 策略下命中缓存的任务应被跳过（覆盖 line 268-274）。"""
+    runs: List[str] = []
+
+    async def make(name: str) -> Any:
+        async def fn() -> str:
+            runs.append(name)
+            return name
+
+        return fn()
+
+    # 用闭包制造可重复调用的 async 函数
+    async def a() -> str:
+        runs.append("a")
+        return "a"
+
+    async def b(a: str) -> str:
+        runs.append("b")
+        return a + "b"
+
+    graph = px.Graph.from_specs(
+        [
+            px.TaskSpec("a", a),
+            px.TaskSpec("b", b, ("a",)),
+        ]
+    )
+    backend = px.MemoryBackend()
+    px.run(graph, strategy="async", state=backend)
+    assert runs == ["a", "b"]
+    px.run(graph, strategy="async", state=backend)
+    assert runs == ["a", "b"]
+
+
+def test_async_all_cached_layer() -> None:
+    """async 整层全部命中缓存（覆盖 line 279 的 if not to_run: return）。"""
+
+    async def a() -> int:
+        return 1
+
+    graph = px.Graph.from_specs([px.TaskSpec("a", a)])
+    backend = px.MemoryBackend()
+    backend.save("a", 77)
+    report = px.run(graph, strategy="async", state=backend)
+    assert report["a"] == 77
+    assert report.result_of("a").status == px.TaskStatus.SKIPPED
+
+
+# ---------------------------------------------------------------------- #
+# 失败后 report.success 标记为 False
+# ---------------------------------------------------------------------- #
+def test_failure_marks_report_unsuccessful() -> None:
+    def boom() -> None:
+        raise ValueError("fail")
+
+    graph = px.Graph.from_specs([px.TaskSpec("a", boom)])
+    with pytest.raises(px.TaskFailedError):
+        px.run(graph, strategy="sequential")
+    # report 在异常前未返回，但若捕获异常则 success 应为 False
+    # 这里验证 run() 抛异常的行为本身
+
+
+# ---------------------------------------------------------------------- #
+# dry_run 各策略
+# ---------------------------------------------------------------------- #
+def test_dry_run_thread(capsys: pytest.CaptureFixture[str]) -> None:
+    graph = px.Graph.from_specs([px.TaskSpec("a", lambda: 1)])  # type: ignore[arg-type]
+    report = px.run(graph, strategy="thread", dry_run=True)
+    assert len(report) == 0
+    assert "Dry run" in capsys.readouterr().out
+
+
+def test_dry_run_async(capsys: pytest.CaptureFixture[str]) -> None:
+    async def a() -> int:
+        return 1
+
+    graph = px.Graph.from_specs([px.TaskSpec("a", a)])
+    report = px.run(graph, strategy="async", dry_run=True)
+    assert len(report) == 0
+    assert "Dry run" in capsys.readouterr().out
+
+
+# ---------------------------------------------------------------------- #
+# 空图运行
+# ---------------------------------------------------------------------- #
+def test_run_empty_graph() -> None:
+    graph = px.Graph()
+    report = px.run(graph, strategy="sequential")
+    assert report.success
+    assert len(report) == 0
diff --git a/tests/test_graph.py b/tests/test_graph.py
index fd6294c..ad268c3 100644
--- a/tests/test_graph.py
+++ b/tests/test_graph.py
@@ -13,11 +13,13 @@ def _fn() -> None:
 
 
 def test_from_specs_builds_graph() -> None:
-    graph = px.Graph.from_specs([
-        px.TaskSpec("a", _fn),
-        px.TaskSpec("b", _fn, ("a",)),
-        px.TaskSpec("c", _fn, ("a", "b")),
-    ])
+    graph = px.Graph.from_specs(
+        [
+            px.TaskSpec("a", _fn),
+            px.TaskSpec("b", _fn, ("a",)),
+            px.TaskSpec("c", _fn, ("a", "b")),
+        ]
+    )
     assert set(graph.names) == {"a", "b", "c"}
     assert graph.dependencies("c") == ("a", "b")
     assert len(graph) == 3
@@ -26,19 +28,23 @@ def test_from_specs_builds_graph() -> None:
 
 def test_from_specs_allows_forward_references() -> None:
     # b depends on a, but a is declared after b — order should not matter.
-    graph = px.Graph.from_specs([
-        px.TaskSpec("b", _fn, ("a",)),
-        px.TaskSpec("a", _fn),
-    ])
+    graph = px.Graph.from_specs(
+        [
+            px.TaskSpec("b", _fn, ("a",)),
+            px.TaskSpec("a", _fn),
+        ]
+    )
     assert graph.layers() == [["a"], ["b"]]
 
 
 def test_duplicate_task_raises() -> None:
     with pytest.raises(DuplicateTaskError):
-        px.Graph.from_specs([
-            px.TaskSpec("a", _fn),
-            px.TaskSpec("a", _fn),
-        ])
+        px.Graph.from_specs(
+            [
+                px.TaskSpec("a", _fn),
+                px.TaskSpec("a", _fn),
+            ]
+        )
 
 
 def test_missing_dependency_raises() -> None:
@@ -50,20 +56,24 @@ def test_missing_dependency_raises() -> None:
 
 def test_cycle_detection() -> None:
     with pytest.raises(CycleError):
-        px.Graph.from_specs([
-            px.TaskSpec("a", _fn, ("c",)),
-            px.TaskSpec("b", _fn, ("a",)),
-            px.TaskSpec("c", _fn, ("b",)),
-        ])
+        px.Graph.from_specs(
+            [
+                px.TaskSpec("a", _fn, ("c",)),
+                px.TaskSpec("b", _fn, ("a",)),
+                px.TaskSpec("c", _fn, ("b",)),
+            ]
+        )
 
 
 def test_layers_grouping() -> None:
-    graph = px.Graph.from_specs([
-        px.TaskSpec("a", _fn),
-        px.TaskSpec("b", _fn),
-        px.TaskSpec("c", _fn, ("a", "b")),
-        px.TaskSpec("d", _fn, ("c",)),
-    ])
+    graph = px.Graph.from_specs(
+        [
+            px.TaskSpec("a", _fn),
+            px.TaskSpec("b", _fn),
+            px.TaskSpec("c", _fn, ("a", "b")),
+            px.TaskSpec("d", _fn, ("c",)),
+        ]
+    )
     layers = graph.layers()
     assert layers == [["a", "b"], ["c"], ["d"]]
 
@@ -74,10 +84,12 @@ def test_self_dependency_rejected() -> None:
 
 
 def test_to_mermaid() -> None:
-    graph = px.Graph.from_specs([
-        px.TaskSpec("a", _fn),
-        px.TaskSpec("b", _fn, ("a",)),
-    ])
+    graph = px.Graph.from_specs(
+        [
+            px.TaskSpec("a", _fn),
+            px.TaskSpec("b", _fn, ("a",)),
+        ]
+    )
     mermaid = graph.to_mermaid()
     assert mermaid.startswith("graph TD")
     assert 'a["a"]' in mermaid
@@ -91,11 +103,13 @@ def test_to_mermaid_invalid_orientation() -> None:
 
 
 def test_subgraph_by_tags() -> None:
-    graph = px.Graph.from_specs([
-        px.TaskSpec("a", _fn, tags=("ingest",)),
-        px.TaskSpec("b", _fn, ("a",), tags=("ingest",)),
-        px.TaskSpec("c", _fn, ("b",), tags=("report",)),
-    ])
+    graph = px.Graph.from_specs(
+        [
+            px.TaskSpec("a", _fn, tags=("ingest",)),
+            px.TaskSpec("b", _fn, ("a",), tags=("ingest",)),
+            px.TaskSpec("c", _fn, ("b",), tags=("report",)),
+        ]
+    )
     sub = graph.subgraph(["ingest"])
     assert set(sub.names) == {"a", "b"}
     # Edge to dropped task c is removed; b no longer waits for anything
@@ -104,11 +118,13 @@ def test_subgraph_by_tags() -> None:
 
 
 def test_subgraph_by_names() -> None:
-    graph = px.Graph.from_specs([
-        px.TaskSpec("a", _fn),
-        px.TaskSpec("b", _fn, ("a",)),
-        px.TaskSpec("c", _fn, ("b",)),
-    ])
+    graph = px.Graph.from_specs(
+        [
+            px.TaskSpec("a", _fn),
+            px.TaskSpec("b", _fn, ("a",)),
+            px.TaskSpec("c", _fn, ("b",)),
+        ]
+    )
     sub = graph.subgraph_by_names(["a", "b"])
     assert set(sub.names) == {"a", "b"}
     # c is dropped, so b's dep on c (none here) — but a->b edge preserved.
@@ -122,10 +138,93 @@ def test_subgraph_by_names_unknown() -> None:
 
 
 def test_describe() -> None:
-    graph = px.Graph.from_specs([
-        px.TaskSpec("a", _fn),
-        px.TaskSpec("b", _fn, ("a",)),
-    ])
+    graph = px.Graph.from_specs(
+        [
+            px.TaskSpec("a", _fn),
+            px.TaskSpec("b", _fn, ("a",)),
+        ]
+    )
     desc = graph.describe()
     assert "Layer 1" in desc
     assert "Layer 2" in desc
+
+
+# ---------------------------------------------------------------------- #
+# 增量 add API 与其他访问器
+# ---------------------------------------------------------------------- #
+def test_add_chains_and_validates() -> None:
+    """add() 应返回 self 以支持链式调用，并即时校验。"""
+    graph = px.Graph()
+    ret = graph.add(px.TaskSpec("a", _fn))
+    assert ret is graph
+    assert "a" in graph
+    # 缺失依赖应即时报错
+    with pytest.raises(MissingDependencyError):
+        graph.add(px.TaskSpec("b", _fn, ("missing",)))
+
+
+def test_add_duplicate_raises() -> None:
+    graph = px.Graph()
+    graph.add(px.TaskSpec("a", _fn))
+    with pytest.raises(DuplicateTaskError):
+        graph.add(px.TaskSpec("a", _fn))
+
+
+def test_all_specs_returns_view() -> None:
+    graph = px.Graph.from_specs([px.TaskSpec("a", _fn)])
+    view = graph.all_specs()
+    assert set(view.keys()) == {"a"}
+    # 返回的是只读视图，修改不影响内部
+    assert view is graph.all_specs() or view == graph.all_specs()
+
+
+def test_spec_accessor() -> None:
+    graph = px.Graph.from_specs([px.TaskSpec("a", _fn)])
+    assert graph.spec("a").name == "a"
+    with pytest.raises(KeyError):
+        graph.spec("missing")
+
+
+def test_dependencies_accessor() -> None:
+    graph = px.Graph.from_specs(
+        [
+            px.TaskSpec("a", _fn),
+            px.TaskSpec("b", _fn, ("a",)),
+        ]
+    )
+    assert graph.dependencies("a") == ()
+    assert graph.dependencies("b") == ("a",)
+
+
+def test_repr() -> None:
+    graph = px.Graph.from_specs([px.TaskSpec("a", _fn)])
+    assert repr(graph) == "Graph(tasks=1)"
+
+
+def test_empty_graph_layers() -> None:
+    """空图的 layers() 应返回空列表。"""
+    graph = px.Graph()
+    assert graph.layers() == []
+    assert graph.to_mermaid() == "graph TD\n"
+
+
+def test_subgraph_preserves_metadata() -> None:
+    """子图应保留原任务的 retries/timeout/tags 等元数据。"""
+    graph = px.Graph.from_specs(
+        [
+            px.TaskSpec("a", _fn, tags=("x",), retries=3, timeout=5.0),
+            px.TaskSpec("b", _fn, ("a",), tags=("y",)),
+        ]
+    )
+    sub = graph.subgraph(["x"])
+    spec = sub.spec("a")
+    assert spec.retries == 3
+    assert spec.timeout == 5.0
+    assert spec.tags == ("x",)
+
+
+def test_subgraph_by_tags_no_match() -> None:
+    """无匹配 tag 时返回空图。"""
+    graph = px.Graph.from_specs([px.TaskSpec("a", _fn, tags=("x",))])
+    sub = graph.subgraph(["z"])
+    assert len(sub) == 0
diff --git a/tests/test_report.py b/tests/test_report.py
new file mode 100644
index 0000000..a4dc3d6
--- /dev/null
+++ b/tests/test_report.py
@@ -0,0 +1,121 @@
+"""RunReport 测试。"""
+
+from __future__ import annotations
+
+from datetime import datetime
+
+import pyflowx as px
+from pyflowx.task import TaskResult, TaskSpec, TaskStatus
+
+
+def _fn() -> int:
+    return 1
+
+
+def _make_result(
+    name: str = "a",
+    status: TaskStatus = TaskStatus.SUCCESS,
+    value: object = 42,
+    error: object = None,
+    duration: float = 0.5,
+    attempts: int = 1,
+) -> TaskResult[object]:
+    spec: TaskSpec[object] = TaskSpec(name, _fn)  # type: ignore[arg-type]
+    start = datetime(2024, 1, 1, 0, 0, 0)
+    # 用 timedelta 精确表达秒数，避免 int() 截断小数
+    from datetime import timedelta
+
+    end = start + timedelta(seconds=duration) if duration else None
+    return TaskResult(
+        spec=spec,
+        status=status,
+        value=value,  # type: ignore[arg-type]
+        error=error,  # type: ignore[arg-type]
+        attempts=attempts,
+        started_at=start,
+        finished_at=end,
+    )
+
+
+def test_getitem_returns_value() -> None:
+    report = px.RunReport()
+    report.results["a"] = _make_result("a", value=7)
+    assert report["a"] == 7
+
+
+def test_result_of_returns_full_result() -> None:
+    report = px.RunReport()
+    r = _make_result("a")
+    report.results["a"] = r
+    assert report.result_of("a") is r
+
+
+def test_contains() -> None:
+    report = px.RunReport()
+    report.results["a"] = _make_result("a")
+    assert "a" in report
+    assert "b" not in report
+
+
+def test_iter_and_len() -> None:
+    report = px.RunReport()
+    report.results["a"] = _make_result("a")
+    report.results["b"] = _make_result("b")
+    assert list(report) == ["a", "b"]
+    assert len(report) == 2
+
+
+def test_summary_success() -> None:
+    report = px.RunReport()
+    report.results["a"] = _make_result("a", status=TaskStatus.SUCCESS, duration=1.0)
+    report.results["b"] = _make_result("b", status=TaskStatus.SKIPPED, duration=0.0)
+    s = report.summary()
+    assert s["success"] is True
+    assert s["total_tasks"] == 2
+    assert s["by_status"] == {"success": 1, "skipped": 1}
+    assert s["total_duration_seconds"] == 1.0
+
+
+def test_summary_with_none_duration() -> None:
+    """未开始/未结束的任务 duration 为 None，不应计入总时长。"""
+    report = px.RunReport()
+    spec: TaskSpec[object] = TaskSpec("a", _fn)  # type: ignore[arg-type]
+    report.results["a"] = TaskResult(spec=spec, status=TaskStatus.FAILED)
+    s = report.summary()
+    assert s["total_duration_seconds"] == 0.0
+
+
+def test_failed_tasks() -> None:
+    report = px.RunReport()
+    report.results["a"] = _make_result("a", status=TaskStatus.SUCCESS)
+    report.results["b"] = _make_result(
+        "b", status=TaskStatus.FAILED, error=ValueError("x")
+    )
+    assert report.failed_tasks() == ["b"]
+
+
+def test_describe_success() -> None:
+    report = px.RunReport()
+    report.results["a"] = _make_result("a", status=TaskStatus.SUCCESS, duration=0.5)
+    desc = report.describe()
+    assert "RunReport(success=True)" in desc
+    assert "a: success" in desc
+    assert "0.500s" in desc
+
+
+def test_describe_with_error() -> None:
+    report = px.RunReport(success=False)
+    report.results["a"] = _make_result(
+        "a", status=TaskStatus.FAILED, error=ValueError("boom"), duration=0.1
+    )
+    desc = report.describe()
+    assert "success=False" in desc
+    assert "error=ValueError" in desc
+
+
+def test_describe_no_duration() -> None:
+    report = px.RunReport()
+    spec: TaskSpec[object] = TaskSpec("a", _fn)  # type: ignore[arg-type]
+    report.results["a"] = TaskResult(spec=spec, status=TaskStatus.PENDING)
+    desc = report.describe()
+    assert "-" in desc  # duration 显示为 "-"
diff --git a/tests/test_storage.py b/tests/test_storage.py
new file mode 100644
index 0000000..9506912
--- /dev/null
+++ b/tests/test_storage.py
@@ -0,0 +1,162 @@
+"""状态后端测试。"""
+
+from __future__ import annotations
+
+import json
+import os
+import tempfile
+from typing import Any
+
+import pytest
+
+from pyflowx.errors import StorageError
+from pyflowx.storage import JSONBackend, MemoryBackend, StateBackend, resolve_backend
+
+
+# ---------------------------------------------------------------------- #
+# MemoryBackend
+# ---------------------------------------------------------------------- #
+def test_memory_backend_lifecycle() -> None:
+    b = MemoryBackend()
+    assert not b.has("a")
+    b.save("a", 1)
+    assert b.has("a")
+    assert b.get("a") == 1
+    assert dict(b.load()) == {"a": 1}
+    b.clear()
+    assert not b.has("a")
+    assert dict(b.load()) == {}
+
+
+def test_memory_backend_get_missing_raises() -> None:
+    b = MemoryBackend()
+    with pytest.raises(KeyError):
+        b.get("nope")
+
+
+# ---------------------------------------------------------------------- #
+# JSONBackend
+# ---------------------------------------------------------------------- #
+def test_json_backend_save_and_load() -> None:
+    with tempfile.TemporaryDirectory() as tmp:
+        path = os.path.join(tmp, "state.json")
+        b = JSONBackend(path)
+        b.save("a", {"x": 1})
+        b.save("b", [1, 2, 3])
+        # 重新打开应读到已保存内容
+        b2 = JSONBackend(path)
+        assert b2.has("a")
+        assert b2.get("a") == {"x": 1}
+        assert b2.get("b") == [1, 2, 3]
+        assert dict(b2.load()) == {"a": {"x": 1}, "b": [1, 2, 3]}
+
+
+def test_json_backend_clear() -> None:
+    with tempfile.TemporaryDirectory() as tmp:
+        path = os.path.join(tmp, "state.json")
+        b = JSONBackend(path)
+        b.save("a", 1)
+        b.clear()
+        assert not b.has("a")
+        # 文件应被写入空 dict
+        with open(path, "r", encoding="utf-8") as fh:
+            assert json.load(fh) == {}
+
+
+def test_json_backend_nonexistent_file_starts_empty() -> None:
+    """文件不存在时应正常初始化为空。"""
+    with tempfile.TemporaryDirectory() as tmp:
+        path = os.path.join(tmp, "absent.json")
+        b = JSONBackend(path)
+        assert dict(b.load()) == {}
+        assert not b.has("anything")
+
+
+def test_json_backend_non_serialisable_raises() -> None:
+    """不可 JSON 序列化的值应抛 StorageError，且不污染内存状态。"""
+    with tempfile.TemporaryDirectory() as tmp:
+        path = os.path.join(tmp, "state.json")
+        b = JSONBackend(path)
+        with pytest.raises(StorageError):
+            b.save("a", object())  # object() 不可序列化
+        assert not b.has("a")
+
+
+def test_json_backend_flush_type_error(monkeypatch: pytest.MonkeyPatch) -> None:
+    """_flush 时 json.dump 抛 TypeError 应转为 StorageError（覆盖 line 105-106）。
+
+    通过 monkeypatch 让 json.dump 在写入文件时抛 TypeError，模拟值通过
+    save 的 dumps 校验但在 dump 到文件句柄时失败（如自定义对象的边缘情况）。
+    """
+    import json as _json
+
+    with tempfile.TemporaryDirectory() as tmp:
+        path = os.path.join(tmp, "state.json")
+        b = JSONBackend(path)
+
+        original_dump = _json.dump
+
+        def flaky_dump(*args: Any, **kwargs: Any) -> None:
+            raise TypeError("simulated flush failure")
+
+        monkeypatch.setattr(_json, "dump", flaky_dump)
+        with pytest.raises(StorageError, match="cannot write"):
+            b.save("a", 1)
+        # 恢复以便后续测试不受影响
+        monkeypatch.setattr(_json, "dump", original_dump)
+
+
+def test_json_backend_flush_os_error(monkeypatch: pytest.MonkeyPatch) -> None:
+    """_flush 时 OSError 应转为 StorageError。"""
+    with tempfile.TemporaryDirectory() as tmp:
+        path = os.path.join(tmp, "state.json")
+        b = JSONBackend(path)
+
+        original_replace = os.replace
+
+        def fail_replace(*args: Any, **kwargs: Any) -> None:
+            raise OSError("simulated os.replace failure")
+
+        monkeypatch.setattr(os, "replace", fail_replace)
+        with pytest.raises(StorageError, match="cannot write"):
+            b.save("a", 1)
+        monkeypatch.setattr(os, "replace", original_replace)
+
+
+def test_json_backend_corrupt_file_raises() -> None:
+    """损坏的 JSON 文件应抛 StorageError。"""
+    with tempfile.TemporaryDirectory() as tmp:
+        path = os.path.join(tmp, "state.json")
+        with open(path, "w", encoding="utf-8") as fh:
+            fh.write("{not valid json")
+        with pytest.raises(StorageError):
+            JSONBackend(path)
+
+
+def test_json_backend_non_dict_content_ignored() -> None:
+    """文件内容是合法 JSON 但非 dict 时应被忽略（保持空）。"""
+    with tempfile.TemporaryDirectory() as tmp:
+        path = os.path.join(tmp, "state.json")
+        with open(path, "w", encoding="utf-8") as fh:
+            json.dump([1, 2, 3], fh)  # list 而非 dict
+        b = JSONBackend(path)
+        assert dict(b.load()) == {}
+
+
+# ---------------------------------------------------------------------- #
+# resolve_backend
+# ---------------------------------------------------------------------- #
+def test_resolve_backend_returns_input() -> None:
+    b = MemoryBackend()
+    assert resolve_backend(b) is b
+
+
+def test_resolve_backend_creates_memory_when_none() -> None:
+    b = resolve_backend(None)
+    assert isinstance(b, MemoryBackend)
+
+
+def test_state_backend_is_abstract() -> None:
+    """StateBackend 是 ABC，不能直接实例化。"""
+    with pytest.raises(TypeError):
+        StateBackend()  # type: ignore[abstract]
diff --git a/tests/test_task.py b/tests/test_task.py
new file mode 100644
index 0000000..5a38878
--- /dev/null
+++ b/tests/test_task.py
@@ -0,0 +1,66 @@
+"""TaskSpec / TaskResult 数据结构测试。"""
+
+from __future__ import annotations
+
+from datetime import datetime
+
+import pytest
+
+import pyflowx as px
+from pyflowx.task import TaskResult, TaskSpec, TaskStatus
+
+
+def _fn() -> None:
+    return None
+
+
+def test_spec_empty_name_rejected() -> None:
+    with pytest.raises(ValueError, match="non-empty"):
+        TaskSpec("", _fn)
+
+
+def test_spec_negative_retries_rejected() -> None:
+    with pytest.raises(ValueError, match="retries"):
+        TaskSpec("a", _fn, retries=-1)
+
+
+def test_spec_zero_timeout_rejected() -> None:
+    with pytest.raises(ValueError, match="timeout"):
+        TaskSpec("a", _fn, timeout=0)
+
+
+def test_spec_self_dependency_rejected() -> None:
+    with pytest.raises(ValueError, match="depend on itself"):
+        TaskSpec("a", _fn, depends_on=("a",))
+
+
+def test_task_result_duration_none_when_not_started() -> None:
+    spec: TaskSpec[None] = TaskSpec("a", _fn)
+    result: TaskResult[None] = TaskResult(spec=spec)
+    assert result.duration is None
+
+
+def test_task_result_duration_when_partial() -> None:
+    spec: TaskSpec[None] = TaskSpec("a", _fn)
+    result: TaskResult[None] = TaskResult(spec=spec, started_at=datetime.now())
+    # started_at 已设但 finished_at 未设 -> None
+    assert result.duration is None
+
+
+def test_task_result_duration_computed() -> None:
+    spec: TaskSpec[None] = TaskSpec("a", _fn)
+    start = datetime(2024, 1, 1, 0, 0, 0)
+    end = datetime(2024, 1, 1, 0, 0, 5)
+    result: TaskResult[None] = TaskResult(
+        spec=spec, started_at=start, finished_at=end
+    )
+    assert result.duration == 5.0
+
+
+def test_task_result_default_status() -> None:
+    spec: TaskSpec[None] = TaskSpec("a", _fn)
+    result: TaskResult[None] = TaskResult(spec=spec)
+    assert result.status == TaskStatus.PENDING
+    assert result.value is None
+    assert result.error is None
+    assert result.attempts == 0