feat(literature): add literatures (#63)

BinWang28 · web-flow · commit 6cd8689e69b1 · 2025-10-07T19:18:19.000+08:00
* update literature

* to pass lint
diff --git a/docs/mkdocs/docs/all_about_agents.md b/docs/mkdocs/docs/all_about_agents.md
@@ -7,104 +7,230 @@ Welcome to our comprehensive resource collection for AI agents. This page curate
 ## Table of Contents
 
 !!! abstract "Resource Categories"
-    1. [Agent Frameworks](#agent-frameworks)
-    2. [Agent Memory](#agent-memory)
-    3. [Papers](#papers)
-    4. [Evaluation](#evaluation)
+    1. [Agent Papers](#papers)
+    2. [Agent Frameworks](#agent-frameworks)
+    3. [Evaluation](#evaluation)
+    4. [Agent Memory](#agent-memory)
+    5. [Blogs](#blogs)
 
 ---
 
-## Agent Frameworks
-
-!!! info "Popular Agent Development Frameworks"
-    Comprehensive frameworks for building and deploying AI agents across different domains.
-
-- **MiroFlow**: Build, manage, and scale your AI agents with ease
-    - [:material-github: GitHub](https://github.com/MiroMindAI/MiroFlow)
-
-- **Youtu-Agent**: A simple yet powerful agent framework that delivers with open-source models
-    - [:material-github: GitHub](https://github.com/TencentCloudADP/youtu-agent)
-
-- **OpenManus**: No fortress, purely open ground. OpenManus is Coming
-    - [:material-github: GitHub](https://github.com/FoundationAgents/OpenManus)
-
-- **OpenBB Platform**: Financial data platform for analysts, quants and AI agents 
-    - [:material-link: Project](https://github.com/OpenBB-finance/OpenBB)
-
----
-
-## Agent Memory
-
-!!! tip "Memory Systems for Persistent Agent Intelligence"
-    Advanced memory solutions for building agents with long-term context and learning capabilities.
-
-- **Mem0**: Building Production- Ready AI Agents with Scalable Long-Term Memory
-    - [:material-github: GitHub](https://github.com/mem0ai/mem0)
-
-- **memobase**: Profile-Based Long-Term Memory for AI Applications
-    - [:material-github: GitHub](https://github.com/memodb-io/memobase)
-
-- **Memento**: Fine-tuning LLM Agents without Fine-tuning LLMs
-    - [:material-file-document: Paper](https://arxiv.org/abs/2508.16153) · [:material-github: GitHub](https://github.com/Agent-on-the-Fly/Memento)
-
----
-
-
-## Papers
+### Agent Papers
 
 !!! note "Research Papers & Publications"
     Latest research in agent systems, methodologies, and theoretical foundations.
 
+- **WebThinker**: Empowering Large Reasoning Models with Deep Research Capability
+    - [:material-file-document: Paper](https://arxiv.org/abs/2504.21776) · [:material-github: GitHub](https://github.com/sunnynexus/WebThinker)
 - **Profile-Aware Maneuvering**: A Dynamic Multi-Agent System for Robust GAIA Problem Solving by AWorld 
     - [:material-file-document: Paper](https://arxiv.org/abs/2508.09889)
-
 - **AFlow**: Automating Agentic Workflow Generation 
     - [:material-file-document: Paper](https://arxiv.org/abs/2410.10762)
-
 - **AgentFly**: Fine-tuning LLM Agents without Fine-tuning LLMs 
     - [:material-file-document: Paper](https://arxiv.org/abs/2508.16153v2)
-
 - **Throttling Web Agents Using Reasoning Gates**
     - [:material-file-document: Paper](https://arxiv.org/abs/2509.01619)
-
 - **The Landscape of Agentic Reinforcement Learning for LLMs**: A Survey
     - [:material-file-document: Paper](https://arxiv.org/abs/2509.02547)
+- **BrowseMaster**: Towards Scalable Web Browsing via Tool-Augmented Programmatic Agent Pair
+    - [:material-file-document: Paper](https://arxiv.org/abs/2508.09129) · [:material-github: GitHub](https://github.com/sjtu-sai-agents/Browse-Master)
+- **Long Term Memory**: The Foundation of AI Self-Evolution
+    - [:material-file-document: Paper](https://arxiv.org/abs/2410.15665)
+- **DeepResearcher**: Scaling Deep Research via Reinforcement Learning in Real-world Environments
+    - [:material-file-document: Paper](https://arxiv.org/abs/2504.03160) · [:material-github: GitHub](https://github.com/GAIR-NLP/DeepResearcher)
+- **Web-Shepherd**: Advancing PRMs for Reinforcing Web Agents
+    - [:material-file-document: Paper](https://arxiv.org/abs/2505.15277) · [:material-github: GitHub](https://github.com/kyle8581/Web-Shepherd)
+- **SimpleDeepSearcher**: Deep Information Seeking via Web-Powered Reasoning Trajectory Synthesis
+    - [:material-file-document: Paper](https://arxiv.org/abs/2505.16834) · [:material-github: GitHub](https://github.com/RUCAIBox/SimpleDeepSearcher)
+- **Alita**: Generalist Agent Enabling Scalable Agentic Reasoning with Minimal Predefinition and Maximal Self-Evolution
+    - [:material-file-document: Paper](https://arxiv.org/abs/2505.20286) · [:material-github: GitHub](https://github.com/CharlesQ9/Alita)
+- **MCP-Zero**: Active Tool Discovery for Autonomous LLM Agents
+    - [:material-file-document: Paper](https://arxiv.org/abs/2506.01056) · [:material-github: GitHub](https://github.com/xfey/MCP-Zero)
+- **AgentOrchestra**: Orchestrating Hierarchical Multi-Agent Intelligence with the Tool-Environment-Agent(TEA) Protocol
+    - [:material-file-document: Paper](https://arxiv.org/abs/2506.12508) · [:material-github: GitHub](https://github.com/SkyworkAI/DeepResearchAgent)
+- **Deep Research Agents**: A Systematic Examination And Roadmap
+    - [:material-file-document: Paper](https://arxiv.org/abs/2506.18096) · [:material-github: GitHub](https://github.com/ai-agents-2030/awesome-deep-research-agent)
+- **SciMaster**: Towards General-Purpose Scientific AI Agents, Part I. X-Master as Foundation: Can We Lead on Humanity's Last Exam?
+    - [:material-file-document: Paper](https://arxiv.org/abs/2507.05241) · [:material-github: GitHub](https://github.com/sjtu-sai-agents/X-Master)
+- **Deep Researcher with Test-Time Diffusion**: Enhancing research capabilities through diffusion-based test-time adaptation
+    - [:material-file-document: Paper](https://arxiv.org/abs/2507.16075)
+- **Multi-Agent Tool-Integrated Policy Optimization**: Enhancing multi-agent systems through integrated tool usage and policy optimization
+    - [:material-file-document: Paper](https://arxiv.org/abs/2510.04678)
+- **WALT**: Web Agents that Learn Tools
+    - [:material-file-document: Paper](https://arxiv.org/abs/2510.01524)
+- **Learning to Route**: A Rule-Driven Agent Framework for Hybrid-Source Retrieval-Augmented Generation
+    - [:material-file-document: Paper](https://arxiv.org/abs/2510.02388)
+- **SurveyBench**: Can LLM(-Agents) Write Academic Surveys that Align with Reader Needs?
+    - [:material-file-document: Paper](https://arxiv.org/abs/2510.03120)
+- **FocusAgent**: Simple Yet Effective Ways of Trimming the Large Context of Web Agents
+    - [:material-file-document: Paper](https://arxiv.org/pdf/2510.03204)
+- **LLM-Based Data Science Agents**: A Survey of Capabilities, Challenges, and Future Directions
+    - [:material-file-document: Paper](https://arxiv.org/pdf/2510.04023)
+- **Agentic Context Engineering**: Evolving Contexts for Self-Improving Language Models
+    - [:material-file-document: Paper](https://arxiv.org/abs/2510.04618)
+- **MARS**: Optimizing Dual-System Deep Research via Multi-Agent Reinforcement Learning
+    - [:material-file-document: Paper](https://arxiv.org/abs/2510.04935)
+- **QuantAgents**: Towards Multi-agent Financial System via Simulated Trading
+    - [:material-file-document: Paper](https://arxiv.org/abs/2510.04643) · [:material-link: Project](https://quantagents.github.io/)
+- **Small Language Models for Agentic Systems**: A Survey of Architectures, Capabilities, and Deployment Trade-offs
+    - [:material-file-document: Paper](https://arxiv.org/abs/2510.03847)
+- **Open Agent Specification (Agent Spec)**: Technical Report
+    - [:material-file-document: Paper](https://arxiv.org/abs/2510.04173v1)
+- **AudioToolAgent**: An Agentic Framework for Audio-Language Models
+    - [:material-file-document: Paper](https://arxiv.org/abs/2510.02995)
+- **ThinkBrake**: Mitigating Overthinking in Tool Reasoning
+    - [:material-file-document: Paper](https://arxiv.org/abs/2510.00546)
+- **TOUCAN**: Synthesizing 1.5M Tool-Agentic Trajectories from Real Environments
+    - [:material-file-document: Paper](https://arxiv.org/abs/2510.01179)
+- **ToolTweak**: An Attack on Tool Selection in LLM-Based Agents
+    - [:material-file-document: Paper](https://arxiv.org/abs/2510.02554)
+- **ToolBrain**: A Flexible RL Framework for Agentic Tools
+    - [:material-file-document: Paper](https://arxiv.org/abs/2510.00023)
+- **TUMIX**: Multi-Agent Test-Time Scaling with Tool-Use Mixture
+    - [:material-file-document: Paper](https://arxiv.org/abs/2510.01279)
+
 
 ---
 
+### Agent Frameworks
+
+!!! info "Popular Agent Development Frameworks"
+    Comprehensive frameworks for building and deploying AI agents across different domains.
+
+- **MiroFlow**: Build, manage, and scale your AI agents with ease
+    - [:material-github: GitHub](https://github.com/MiroMindAI/MiroFlow)
+- **Youtu-Agent**: A simple yet powerful agent framework that delivers with open-source models
+    - [:material-github: GitHub](https://github.com/TencentCloudADP/youtu-agent)
+- **OpenManus**: No fortress, purely open ground. OpenManus is Coming
+    - [:material-github: GitHub](https://github.com/FoundationAgents/OpenManus)
+- **OpenBB Platform**: Financial data platform for analysts, quants and AI agents 
+    - [:material-link: Project](https://github.com/OpenBB-finance/OpenBB)
+- **TradingAgents**: Multi-Agents LLM Financial Trading Framework
+    - [:material-file-document: Paper](https://arxiv.org/abs/2412.20138) · [:material-github: GitHub](https://github.com/TauricResearch/TradingAgents)
+- **JoyAgent-JDGenie**: Technical Report on the GAIA
+    - [:material-file-document: Paper](https://arxiv.org/abs/2510.00510) · [:material-github: GitHub](https://github.com/jd-opensource/joyagent-jdgenie)
+
 
 
-## Evaluation
+---
+
+### Evaluation
 
 !!! success "Benchmarks & Evaluation Frameworks"
     Comprehensive evaluation tools and benchmarks for measuring agent performance across various tasks.
 
 - **LiveMCP-101**: Stress Testing and Diagnosing MCP-enabled Agents on Challenging Queries 
     - [:material-file-document: Paper](https://arxiv.org/abs/2508.15760)
-
 - **BrowseComp-Plus**: A More Fair and Transparent Evaluation Benchmark of Deep-Research Agent 
     - [:material-file-document: Paper](https://arxiv.org/abs/2508.06600)
-
 - **HotpotQA**: A Dataset for Diverse, Explainable Multi-hop Question Answering
     - [:material-file-document: Paper](https://arxiv.org/abs/1809.09600)
-
 - **GAIA**: a benchmark for General AI Assistants 
     - [:material-file-document: Paper](https://arxiv.org/abs/2311.12983) · [:material-trophy: Leaderboard](https://huggingface.co/spaces/gaia-benchmark/leaderboard)
-
 - **xbench**: Tracking Agents Productivity Scaling with Profession-Aligned Real-World Evaluations 
     - [:material-file-document: Paper](https://arxiv.org/abs/2506.13651)
-
 - **MCP-Universe**: Benchmarking Large Language Models with Real-World Model Context Protocol Servers 
     - [:material-file-document: Paper](https://arxiv.org/abs/2508.14704)
-
 - **FutureX**: An Advanced Live Benchmark for LLM Agents in Future Prediction 
     - [:material-file-document: Paper](https://arxiv.org/abs/2508.11987)
-
 - **Terminal-Bench**: the benchmark for testing AI agents in real terminal environments 
     - [:material-github: GitHub](https://github.com/laude-institute/terminal-bench)
-
 - **Gaia2 and ARE**: Empowering the Community to Evaluate Agents
     - [:material-file-document: Blog Post](https://huggingface.co/blog/gaia2)
+- **GPQA**: A Graduate-Level Google-Proof Q&A Benchmark
+    - [:material-file-document: Paper](https://arxiv.org/abs/2311.12022) · [:material-github: GitHub](https://github.com/idavidrein/gpqa/)
+- **WebWalkerQA**: WebWalker: Benchmarking LLMs in Web Traversal
+    - [:material-file-document: Paper](https://arxiv.org/abs/2501.07572) · [:material-github: GitHub](https://github.com/Alibaba-NLP/DeepResearch) · [:material-trophy: Leaderboard](https://huggingface.co/spaces/callanwu/WebWalkerQALeaderboard)
+- **HLE**: Humanity's Last Exam
+    - [:material-file-document: Paper](https://arxiv.org/abs/2501.14249) · [:material-link: Website](https://lastexam.ai/)
+- **BFCL**: Berkeley Function Calling Leaderboard
+    - [:material-github: GitHub](https://github.com/ShishirPatil/gorilla) · [:material-trophy: Leaderboard](https://gorilla.cs.berkeley.edu/leaderboard.html)
+- **When2Call**: When (not) to Call Tools
+    - [:material-file-document: Paper](https://arxiv.org/abs/2504.18851) · [:material-github: GitHub](https://github.com/NVIDIA/When2Call)
+- **ToolSandbox**: A Stateful, Conversational, Interactive Evaluation Benchmark for LLM Tool Use Capabilities
+    - [:material-file-document: Paper](https://arxiv.org/abs/2408.04682) · [:material-github: GitHub](https://github.com/apple/ToolSandbox)
+- **ToolBeHonest**: A Multi-level Hallucination Diagnostic Benchmark for Tool-Augmented Large Language Models
+    - [:material-file-document: Paper](https://arxiv.org/abs/2406.20015) · [:material-github: GitHub](https://github.com/ToolBeHonest/ToolBeHonest)
+- **SuperGPQA**: Scaling LLM Evaluation across 285 Graduate Disciplines
+    - [:material-file-document: Paper](https://arxiv.org/abs/2502.14739) · [:material-link: Website](https://supergpqa.github.io/)
+- **Terminal-Bench**: A benchmark for testing AI agents in terminal environments
+    - [:material-trophy: Leaderboard](https://www.tbench.ai/leaderboard) · [:material-link: Website](https://www.tbench.ai/)
+- **τ-bench**: A Benchmark for Tool-Agent-User Interaction in Real-World Domains
+    - [:material-file-document: Paper](https://arxiv.org/abs/2406.12045) · [:material-github: GitHub](https://github.com/sierra-research/tau-bench)
+- **τ2-Bench**: Evaluating Conversational Agents in a Dual-Control Environment
+    - [:material-file-document: Paper](https://arxiv.org/abs/2506.07982) · [:material-github: GitHub](https://github.com/sierra-research/tau2-bench)
+- **Deep Research Bench**: Evaluating AI Web Research Agents
+    - [:material-file-document: Paper](https://arxiv.org/abs/2506.06287) · [:material-link: Website](https://drb.futuresearch.ai/)
+- **Beyond the Final Answer**: Evaluating the Reasoning Trajectories of Tool-Augmented Agents
+    - [:material-file-document: Paper](https://arxiv.org/abs/2510.02837)
+- **TRAJECT-Bench**: A Trajectory-Aware Benchmark for Evaluating Agentic Tool Use
+    - [:material-file-document: Paper](https://arxiv.org/abs/2510.04550)
+- **ARC-AGI**: The General Intelligence Benchmark
+    - [:material-link: Website](https://arcprize.org/arc-agi)
+
+
+---
+
+### Agent Memory
+
+!!! tip "Memory Systems for Persistent Agent Intelligence"
+    Advanced memory solutions for building agents with long-term context and learning capabilities.
+
+- **Mem0**: Building Production- Ready AI Agents with Scalable Long-Term Memory
+    - [:material-github: GitHub](https://github.com/mem0ai/mem0)
+- **memobase**: Profile-Based Long-Term Memory for AI Applications
+    - [:material-github: GitHub](https://github.com/memodb-io/memobase)
+- **Memento**: Fine-tuning LLM Agents without Fine-tuning LLMs
+    - [:material-file-document: Paper](https://arxiv.org/abs/2508.16153) · [:material-github: GitHub](https://github.com/Agent-on-the-Fly/Memento)
+- **MEMTRACK**: Evaluating Long-Term Memory and State Tracking in Multi-Platform Dynamic Agent Environments
+    - [:material-file-document: Paper](https://arxiv.org/abs/2510.01353)
+- **A-MEM**: Agentic Memory for LLM Agents
+    - [:material-file-document: Paper](https://arxiv.org/abs/2502.12110) · [:material-github: GitHub](https://github.com/WujiangXu/A-mem)
+- **MemoryOS**: Memory OS of AI Agent
+    - [:material-file-document: Paper](https://arxiv.org/abs/2506.06326) · [:material-github: GitHub](https://github.com/BAI-LAB/MemoryOS)
+- **Memory-R1**: Enhancing Large Language Model Agents to Manage and Utilize Memories via Reinforcement Learning
+    - [:material-file-document: Paper](https://arxiv.org/abs/2508.19828)
+- **HippoRAG**: Neurobiologically Inspired Long-Term Memory for Large Language Models
+    - [:material-file-document: Paper](https://arxiv.org/abs/2405.14831) · [:material-github: GitHub](https://github.com/OSU-NLP-Group/HippoRAG)
+- **MaxKB**: Open-source platform for building enterprise-grade agents
+    - [:material-github: GitHub](https://github.com/1Panel-dev/MaxKB)
+- **MemAgent**: Reshaping Long-Context LLM with Multi-Conv RL-based Memory Agent
+    - [:material-file-document: Paper](https://arxiv.org/abs/2507.02259) · [:material-link: Website](https://memagent-sialab.github.io/)
+- **LEGOMem**: Modular Procedural Memory for Multi-agent LLM Systems for Workflow Automation
+    - [:material-file-document: Paper](https://arxiv.org/abs/2510.04851)
+- **Memp**: Exploring Agent Procedural Memory
+    - [:material-file-document: Paper](https://www.arxiv.org/abs/2508.06433)
+- **MIRIX**: Multi-Agent Memory System for LLM-Based Agents
+    - [:material-file-document: Paper](https://arxiv.org/abs/2507.07957) · [:material-link: Website](https://mirix.io/)
+- **A-MemGuard**: A Proactive Defense Framework for LLM-Based Agent Memory
+    - [:material-file-document: Paper](https://www.arxiv.org/abs/2510.02373)
+
+
+---
+
+## Blogs
+
+!!! info "Blog Posts & Tutorials"
+    Curated collection of blog posts, tutorials, and articles about AI agents from various sources and languages.
+
+#### General Blogs
+
+- **ChatGPT Agent**: Introducing ChatGPT Agent
+    - [:material-link: Blog Post](https://openai.com/index/introducing-chatgpt-agent/) · OpenAI's latest agent capabilities and features
+
+- **Tongyi DeepResearch**: Deep Research Agent for Complex Tasks
+    - [:material-link: Blog Post](https://tongyi-agent.github.io/blog/introducing-tongyi-deep-research/) · Alibaba's advanced research agent system
+
+#### Chinese Blogs
+
+!!! quote "中文博客与资源"
+    精选的中文AI智能体相关博客文章、教程和资源，帮助中文用户更好地理解和应用智能体技术。
+
+- **17个主流 Agent 框架快速对比**
+    - [:material-link: 博客链接](https://zhuanlan.zhihu.com/p/1957319210951746186) · 知乎专栏文章，对比分析主流智能体框架
+
+- **通义 DeepResearch**
+    - [:material-link: Blog Post](https://tongyi-agent.github.io/zh/blog/introducing-tongyi-deep-research/) · 阿里巴巴通义智能体深度研究系统介绍
 
 ---
 
diff --git a/docs/mkdocs/docs/openai-gpt.md b/docs/mkdocs/docs/openai-gpt.md
@@ -6,14 +6,14 @@ OpenAI's latest models including GPT-5, GPT-4o and advanced reasoning models wit
 
 `GPT5OpenAIClient`
 
-## Environment Setup
+### Environment Setup
 
 ```bash title="Environment Variables"
 export OPENAI_API_KEY="your-openai-key"
 export OPENAI_BASE_URL="https://api.openai.com/v1"  # optional
 ```
 
-## Configuration
+### Configuration
 
 ```yaml title="Agent Configuration"
 main_agent:
@@ -31,7 +31,7 @@ main_agent:
     openai_base_url: "${oc.env:OPENAI_BASE_URL,https://api.openai.com/v1}"
 ```
 
-## Usage
+### Usage
 
 ```bash title="Example Command"
 # Create custom OpenAI config
@@ -43,14 +43,14 @@ uv run main.py trace --config_file_name=your_config_file \
 
 `GPTOpenAIClient`
 
-## Environment Setup
+### Environment Setup
 
 ```bash title="Environment Variables"
 export OPENAI_API_KEY="your-openai-key"
 export OPENAI_BASE_URL="https://api.openai.com/v1"  # optional
 ```
 
-## Configuration
+### Configuration
 
 ```yaml title="Agent Configuration"
 main_agent:
@@ -61,7 +61,7 @@ main_agent:
     openai_base_url: "${oc.env:OPENAI_BASE_URL,https://api.openai.com/v1}"
 ```
 
-## Usage
+### Usage
 
 ```bash title="Example Command"
 # Create custom OpenAI config
diff --git a/src/llm/providers/claude_openrouter_client.py b/src/llm/providers/claude_openrouter_client.py
@@ -411,4 +411,4 @@ def _apply_cache_control(self, messages):
             else:
                 # Other messages add directly
                 cached_messages.append(turn)
-        return list(reversed(cached_messages))
+        return list(reversed(cached_messages))
diff --git a/src/llm/providers/gpt5_openai_client.py b/src/llm/providers/gpt5_openai_client.py
@@ -132,7 +132,7 @@ async def _create_message(
                 extra_body["min_p"] = self.min_p
             if self.repetition_penalty != 1.0:
                 extra_body["repetition_penalty"] = self.repetition_penalty
-            
+
             assert self.model_name in ["gpt-5-2025-08-07", "gpt-5"]
             params = {
                 "model": self.model_name,
diff --git a/src/llm/providers/gpt_openai_client.py b/src/llm/providers/gpt_openai_client.py
diff --git a/src/tool/mcp_servers/vision_mcp_server.py b/src/tool/mcp_servers/vision_mcp_server.py
