diff --git a/README.md b/README.md
index e2d9dbd..f69d83e 100644
--- a/README.md
+++ b/README.md
@@ -14,7 +14,7 @@
 
 ## Why CNLLM?
 
-CNLLM Python SDK 为中文大模型提供了一个**统一的 OpenAI 兼容接口层**与一套**标准化的参数规则和响应格式规范**。
+CNLLM Python 工具包为各类中文大语言模型提供**统一的 OpenAI 兼容接口层**，并配套一系列**增强工具**，大幅简化大模型开发流程。
 
 通过 CNLLM，开发者可以无障碍地在 OpenAI 生态内的 langchain、LlamaIndex、AutoGen、Haystack、DeepEval 等主流大模型应用框架中使用中文大模型；尤其在需要多模型协作的开发和应用场景中，使用 CNLLM 可**显著减少适配解析、功能实现及维护工程量，并有效降低 AI agent 开发中的 Token 消耗**。
 
diff --git a/README_en.md b/README_en.md
index b02ef3d..f75bc0e 100644
--- a/README_en.md
+++ b/README_en.md
@@ -14,7 +14,7 @@
 
 ## Why CNLLM?
 
-CNLLM Python SDK provides Chinese LLMs with a **unified OpenAI-compatible interface layer** and a set of **standardized parameter rules and response format specifications**. 
+The CNLLM Python toolkit offers a **unified OpenAI-compatible interface layer** for all Chinese LLMs, alongside a suite of **enhanced utilities** to streamline LLM development workflows.. 
 
 Through CNLLM, developers can seamlessly use Chinese LLMs in the OpenAI ecosystem — LangChain, LlamaIndex, AutoGen, Haystack, DeepEval and other mainstream large language model application frameworks. Especially in development and application scenarios requiring multi-model collaboration, using CNLLM can **significantly reduce adaptation, parsing, feature implementation, and maintenance workload, and effectively lower token consumption in AI agent development**.
 
diff --git a/pyproject.toml b/pyproject.toml
index ac8de23..78fd810 100644
--- a/pyproject.toml
+++ b/pyproject.toml
@@ -5,7 +5,7 @@ build-backend = "setuptools.build_meta"
 [project]
 name = "cnllm"
 version = "0.9.10"
-description = "Unified Python library for Chinese LLMs, with flexible batch capacity, feedback on vendor-native parameter validation, and structured overview and automated accumulation for streaming."
+description = "Python toolkit for Chinese LLMs, with flexible batch capacity, structured real-time visulization and automated accumulation for streaming, and explicit feedback on vendor-native parameter validation."
 readme = "README_en.md"
 license = "Apache-2.0"
 authors = [