diff --git a/lightrag/lightrag/components/reasoning/chain_of_thought.py b/lightrag/lightrag/components/reasoning/chain_of_thought.py
index 699975e2..432613ca 100644
--- a/lightrag/lightrag/components/reasoning/chain_of_thought.py
+++ b/lightrag/lightrag/components/reasoning/chain_of_thought.py
@@ -1,4 +1,6 @@
 """
+Chain of the thought(CoT) is to mimic a step-by-step thought process for arriving at the answer.
+
 https://arxiv.org/abs/2201.11903, published in Jan, 2023
 
 Chain of the thought(CoT) is to mimic a step-by-step thought process for arriving at the answer. You can achieve it in two ways:
diff --git a/lightrag/lightrag/core/component.py b/lightrag/lightrag/core/component.py
index fd6c64fa..b51b1ae8 100644
--- a/lightrag/lightrag/core/component.py
+++ b/lightrag/lightrag/core/component.py
@@ -1,3 +1,5 @@
+"""Component is to LLM task pipelines what nn.Module is to PyTorch models."""
+
 from collections import OrderedDict, namedtuple
 from typing import (
     Callable,
diff --git a/lightrag/lightrag/core/default_prompt_template.py b/lightrag/lightrag/core/default_prompt_template.py
index 65813ee9..3355a8cf 100644
--- a/lightrag/lightrag/core/default_prompt_template.py
+++ b/lightrag/lightrag/core/default_prompt_template.py
@@ -1,3 +1,8 @@
+"""This is the default system prompt template used in the LightRAG.
+
+Use :ref:`Prompt <core-prompt_builder>` class  to manage it.
+"""
+
 # TODO: potentially make a data class for this
 LIGHTRAG_DEFAULT_PROMPT_ARGS = [
     "task_desc_str",  # task description
@@ -75,5 +80,5 @@
 """
 """This is the default system prompt template used in the LightRAG.
 
-Use :ref:`Prompt<core-prompt_builder>` class  to manage it.
+Use :ref:`Prompt <core-prompt_builder>` class  to manage it.
 """
diff --git a/lightrag/lightrag/core/generator.py b/lightrag/lightrag/core/generator.py
index a111b10f..2b23b74c 100644
--- a/lightrag/lightrag/core/generator.py
+++ b/lightrag/lightrag/core/generator.py
@@ -1,3 +1,6 @@
+"""Generator is a user-facing orchestration component with a simple and unified interface for LLM prediction. 
+
+It is a pipeline that consists of three subcomponents."""
 from typing import Any, Dict, List, Optional, Union
 from copy import deepcopy
 import logging
diff --git a/lightrag/lightrag/core/parameter.py b/lightrag/lightrag/core/parameter.py
index e113e254..43e07b49 100644
--- a/lightrag/lightrag/core/parameter.py
+++ b/lightrag/lightrag/core/parameter.py
@@ -1,3 +1,4 @@
+"""WIP"""
 from typing import Generic, TypeVar, Any
 
 T = TypeVar("T")  # covariant set to False to allow for in-place updates
diff --git a/lightrag/lightrag/core/tokenizer.py b/lightrag/lightrag/core/tokenizer.py
index 1a86c6c4..c42af386 100644
--- a/lightrag/lightrag/core/tokenizer.py
+++ b/lightrag/lightrag/core/tokenizer.py
@@ -1,3 +1,6 @@
+"""
+Tokenizer from tiktoken.
+"""
 import tiktoken
 from typing import List
 
diff --git a/lightrag/lightrag/core/tool_manager.py b/lightrag/lightrag/core/tool_manager.py
index d835c658..62b61e53 100644
--- a/lightrag/lightrag/core/tool_manager.py
+++ b/lightrag/lightrag/core/tool_manager.py
@@ -1,3 +1,6 @@
+"""
+The ToolManager manages a list of tools, context, and all ways to execute functions.
+"""
 from typing import List, Dict, Optional, Any, Callable, Awaitable, Union
 import logging
 from copy import deepcopy
diff --git a/lightrag/lightrag/optim/llm_optimizer.py b/lightrag/lightrag/optim/llm_optimizer.py
index 5a615043..0b8141e0 100644
--- a/lightrag/lightrag/optim/llm_optimizer.py
+++ b/lightrag/lightrag/optim/llm_optimizer.py
@@ -1,5 +1,7 @@
 r"""
-Based and optimized from ORPO llm optimizer: https://arxiv.org/abs/2309.03409
+Based and optimized from ORPO llm optimizer.
+
+https://arxiv.org/abs/2309.03409
 Source code: https://github.com/google-deepmind/opro
 """