mirror of
https://git.mirrors.martin98.com/https://github.com/bytedance/deer-flow
synced 2025-08-19 12:59:11 +08:00
feat: add more examples
This commit is contained in:
He Tao
parent
1195612c47
commit
aaa340787c
18
README.md
18
README.md
@ -88,19 +88,31 @@ The following examples demonstrate the capabilities of lite-deep-researcher:
|
||||
|
||||
### Research Reports
|
||||
|
||||
1. **What is MCP?** - A comprehensive analysis of the term "MCP" across multiple contexts
|
||||
1. **OpenAI Sora Report** - Analysis of OpenAI's Sora AI tool
|
||||
- Discusses features, access, prompt engineering, limitations, and ethical considerations
|
||||
- [View full report](examples/openai_sora_report.md)
|
||||
|
||||
2. **Google's Agent to Agent Protocol Report** - Overview of Google's Agent to Agent (A2A) protocol
|
||||
- Discusses its role in AI agent communication and its relationship with Anthropic's Model Context Protocol (MCP)
|
||||
- [View full report](examples/what_is_agent_to_agent_protocol.md)
|
||||
|
||||
3. **What is MCP?** - A comprehensive analysis of the term "MCP" across multiple contexts
|
||||
- Explores Model Context Protocol in AI, Monocalcium Phosphate in chemistry, and Micro-channel Plate in electronics
|
||||
- [View full report](examples/what_is_mcp.md)
|
||||
|
||||
2. **Bitcoin Price Fluctuations** - Analysis of recent Bitcoin price movements
|
||||
4. **Bitcoin Price Fluctuations** - Analysis of recent Bitcoin price movements
|
||||
- Examines market trends, regulatory influences, and technical indicators
|
||||
- Provides recommendations based on historical data
|
||||
- [View full report](examples/bitcoin_price_fluctuation.md)
|
||||
|
||||
3. **What is LLM?** - An in-depth exploration of Large Language Models
|
||||
5. **What is LLM?** - An in-depth exploration of Large Language Models
|
||||
- Discusses architecture, training, applications, and ethical considerations
|
||||
- [View full report](examples/what_is_llm.md)
|
||||
|
||||
6. **How to Use Claude for Deep Research?** - Best practices and workflows for using Claude in deep research
|
||||
- Covers prompt engineering, data analysis, and integration with other tools
|
||||
- [View full report](examples/how_to_use_claude_deep_research.md)
|
||||
|
||||
To run these examples or create your own research reports, you can use the following commands:
|
||||
|
||||
```bash
|
||||
|
||||
77
examples/how_to_use_claude_deep_research.md
Normal file
77
examples/how_to_use_claude_deep_research.md
Normal file
@ -0,0 +1,77 @@
|
||||
# Deep Research with Claude: Workflows and Best Practices
|
||||
|
||||
## Executive Summary
|
||||
|
||||
This report outlines optimal workflows and best practices for integrating Claude into deep research processes, covering data collection, preprocessing, analysis, and synthesis. It also addresses integration with other tools, validation methods, cost management, collaboration strategies, documentation practices, and relevant case studies. Claude can assist in academic writing and research and should be used to support, not replace, original thought.
|
||||
|
||||
## Key Findings
|
||||
|
||||
* Claude can assist in academic writing and research, but should be used to support, not replace, original thought.
|
||||
* Claude's Project feature allows uploading relevant documents to reduce repetitive context-setting.
|
||||
* The AI has a data analysis tool that can write and run JavaScript code to process data and offer insights.
|
||||
* Claude offers citation tools for verifying sources and ensuring proper formatting.
|
||||
* Haiku is the fastest and most cost-effective model in its intelligence category.
|
||||
* Claude can serve as a virtual teammate to advance work.
|
||||
* Sharing work products created with Claude can improve innovation in product development and research.
|
||||
* Claude can create technical documentation faster while maintaining consistency.
|
||||
* Claude integrates with note-taking, writing, and reference management tools.
|
||||
|
||||
## Detailed Analysis
|
||||
|
||||
### Workflows and Best Practices
|
||||
|
||||
* **Define Research Questions:** Clearly define research questions and areas of focus in initial prompts.
|
||||
* **Structured Data:** Provide relevant data in a structured message.
|
||||
* **Project Feature:** Use Claude's Project feature to upload relevant documents, reducing the need for repetitive context-setting.
|
||||
* [Source: [https://support.anthropic.com/en/articles/9797557-usage-limit-best-practices](https://support.anthropic.com/en/articles/9797557-usage-limit-best-practices)]
|
||||
* **Prompt Engineering:** Employ prompt engineering techniques, such as including "Think step by step," to improve performance.
|
||||
* [Source: [https://aws.amazon.com/blogs/machine-learning/prompt-engineering-techniques-and-best-practices-learn-by-doing-with-anthropics-claude-3-on-amazon-bedrock/](https://aws.amazon.com/blogs/machine-learning/prompt-engineering-techniques-and-best-practices-learn-by-doing-with-anthropics-claude-3-on-amazon-bedrock/)]
|
||||
|
||||
### Data Analysis
|
||||
|
||||
* **Data Analysis Tool:** Utilize Claude’s built-in data analysis tool, which writes and runs JavaScript code to process data and provide insights.
|
||||
* [Source: [https://www.anthropic.com/news/analysis-tool](https://www.anthropic.com/news/analysis-tool), [https://support.anthropic.com/en/articles/10008684-enabling-and-using-the-analysis-tool](https://support.anthropic.com/en/articles/10008684-enabling-and-using-the-analysis-tool)]
|
||||
* **CSV Analysis:** Use the data analysis tool to analyze and visualize data from uploaded CSV files.
|
||||
* [Source: [https://support.anthropic.com/en/articles/10008684-enabling-and-using-the-analysis-tool](https://support.anthropic.com/en/articles/10008684-enabling-and-using-the-analysis-tool)]
|
||||
|
||||
### Validation
|
||||
|
||||
* **Citation Tools:** Utilize Claude's citation tools to verify sources and ensure correct formatting for academic rigor.
|
||||
* [Source: [https://www.yomu.ai/blog/claude-ai-in-academic-writing-and-research-essential-tips-for-optimal-results](https://www.yomu.ai/blog/claude-ai-in-academic-writing-and-research-essential-tips-for-optimal-results)]
|
||||
* **Prompt Sanitization:** Note that the Anthropic API performs basic prompt sanitization and validation.
|
||||
* [Source: [https://docs.anthropic.com/en/api/prompt-validation](https://docs.anthropic.com/en/api/prompt-validation)]
|
||||
|
||||
### Cost Management
|
||||
|
||||
* **Model Selection:** Consider using the Haiku model for cost-effective performance in its intelligence category.
|
||||
* [Source: [https://www.anthropic.com/news/claude-3-family](https://www.anthropic.com/news/claude-3-family)]
|
||||
|
||||
### Collaboration
|
||||
|
||||
* **Virtual Teammate:** Leverage Claude as a virtual teammate to move work forward.
|
||||
* [Source: [https://www.anthropic.com/team](https://www.anthropic.com/team)]
|
||||
* **Shared Work Products:** Share work products co-created with Claude to foster innovation, particularly in product development and research.
|
||||
* [Source: [https://www.anthropic.com/news/projects](https://www.anthropic.com/news/projects)]
|
||||
|
||||
### Documentation
|
||||
|
||||
* **Technical Documentation:** Use Claude to create technical documentation more efficiently and maintain consistency.
|
||||
* [Source: [https://beginswithai.com/how-to-use-claude-ai-to-create-technical-documentation/](https://beginswithai.com/how-to-use-claude-ai-to-create-technical-documentation/)]
|
||||
|
||||
### Integration with Other Tools
|
||||
|
||||
* **Note-Taking and Writing Tools:** Integrate Claude with note-taking and writing tools such as Evernote, OneNote, or Google Docs.
|
||||
* [Source: [https://beginswithai.com/using-claude-for-research/](https://beginswithai.com/using-claude-for-research/)]
|
||||
* **Reference Management Tools:** Work with reference management tools like Zotero, Mendeley, and EndNote.
|
||||
* [Source: [https://beginswithai.com/using-claude-for-research/](https://beginswithai.com/using-claude-for-research/)]
|
||||
* **Platform Integration:** Ensure smooth integration with platforms like Anthropic API and Google Cloud's Vertex AI.
|
||||
* [Source: [https://www.yomu.ai/blog/claude-ai-in-academic-writing-and-research-essential-tips-for-optimal-results](https://www.yomu.ai/blog/claude-ai-in-academic-writing-and-research-essential-tips-for-optimal-results)]
|
||||
|
||||
### Case Studies
|
||||
|
||||
* **Diverse Applications:** Explore case studies that demonstrate the successful use of Claude in various domains, including whale conservation, brand management, cybersecurity, hiring, insurance, code review, customer service, and sales.
|
||||
* [Source: [https://www.anthropic.com/customers](https://www.anthropic.com/customers)]
|
||||
|
||||
## Conclusions and Recommendations
|
||||
|
||||
Claude is a valuable tool for deep research if used strategically. By defining clear research questions, providing structured data, and utilizing Claude's project features, researchers can maximize its potential. The AI's data analysis capabilities, especially with CSV files, offer real-time insights. Validating Claude's outputs through citation tools and careful prompt engineering is essential for accuracy. Collaboration features enhance teamwork, and integrations with other research tools streamline workflows. The case studies show the broad applicability of Claude across different fields, highlighting its versatility and potential impact.
|
||||
128
examples/openai_sora_report.md
Normal file
128
examples/openai_sora_report.md
Normal file
@ -0,0 +1,128 @@
|
||||
# OpenAI Sora Usage Report
|
||||
|
||||
## Key Points
|
||||
|
||||
* Sora is OpenAI's text-to-video model that generates videos from text prompts and can extend existing short videos. It was released publicly for ChatGPT Plus and ChatGPT Pro users in December 2024.
|
||||
* Currently, access to Sora is limited, primarily granted to selected developers, visual artists, designers, and filmmakers for testing and feedback purposes. The API is not yet publicly available.
|
||||
* Sora allows users to generate videos with customizable resolutions up to 1080p and lengths up to 20 seconds, supporting various aspect ratios and the incorporation of user-provided assets.
|
||||
* Sora is capable of generating videos in diverse styles, applying camera angles, motion, and lighting effects, and mimicking realistic or imaginative scenarios based on text prompts.
|
||||
* Limitations include potential inaccuracies in simulating physics, biases, and ethical concerns related to deepfakes and misinformation, which OpenAI is addressing with content moderation and community-driven guidelines.
|
||||
* Geographically, Sora is available in over 150 countries but remains inaccessible in the European Union and the UK due to regulatory challenges and a prioritized rollout to US users.
|
||||
|
||||
---
|
||||
|
||||
## Overview
|
||||
|
||||
OpenAI's Sora is a text-to-video model designed to generate short video clips based on user-provided text prompts. Launched in December 2024, Sora represents a significant advancement in AI-driven content creation, allowing users to bring imaginative scenarios to life through video. However, its release is accompanied by both excitement and concerns regarding its capabilities, limitations, and ethical implications.
|
||||
|
||||
---
|
||||
|
||||
## Detailed Analysis
|
||||
|
||||
### Functionalities and Capabilities
|
||||
|
||||
Sora offers a range of functionalities, including:
|
||||
|
||||
* **Text-to-Video Generation**: Creating realistic and imaginative videos from text prompts.
|
||||
* **Video Editing**: Options for remixing, re-cutting, looping, blending, and storyboarding video content.
|
||||
* **Prompt Interpretation**: Generating videos that mimic real-world scenes or bring to life imaginative scenarios.
|
||||
* **Video Styles and Content**: Generating videos in various styles, from realistic to artistic.
|
||||
|
||||
Sora is capable of applying various camera angles, motion, and lighting effects to the generated videos. Specific camera movements like pan, tilt, dolly, zoom, and more can be directed using detailed prompts.
|
||||
|
||||
### Access and Availability
|
||||
|
||||
Currently, access to Sora is limited. It is primarily available to selected developers, visual artists, designers, and filmmakers for the purpose of testing, gathering feedback, and assessing potential weaknesses and risks. The API is not yet publicly available, and OpenAI has not specified a concrete timeline for broader access. It was released publicly for ChatGPT Plus and ChatGPT Pro users in December 2024.
|
||||
|
||||
Geographical availability is also restricted. While Sora is available in more than 150 countries, it is currently inaccessible in the European Union and the UK due to specific EU regulations regarding AI use and an initial focus on US users.
|
||||
|
||||
### Content Limitations and Restrictions
|
||||
|
||||
Sora has several limitations and restrictions:
|
||||
|
||||
* **Resolution and Length**: Videos can be generated up to 1080p resolution, with lengths up to 20 seconds for ChatGPT Pro users and 10 seconds for ChatGPT Plus users. Lower resolutions, such as 480p, are also available.
|
||||
* **Complexity**: The model sometimes struggles with realistic physics and complex actions over long durations.
|
||||
* **Content Restrictions**: There are age restrictions, allowing only adults (above 18 years) to use the tool, and visual content depicting minors is prohibited. There are limitations in depicting humans; for now, only a small group of selected testers can create human-like videos.
|
||||
* **Biases and Inaccuracies**: Sora may not always understand the entire context of a prompt, leading to inaccurate or irrelevant outputs and potential biases perpetuating stereotypes.
|
||||
|
||||
### Ethical Considerations and Policies
|
||||
|
||||
Sora raises ethical concerns related to the creation of deepfakes and the potential spread of misinformation. OpenAI is aware of these concerns and is implementing policies and safeguards to address them:
|
||||
|
||||
* **Content Moderation**: Features to promote responsible use and prohibit harmful content.
|
||||
* **Community Guidelines**: Community-driven guidelines to ensure Sora responds to cultural diversity.
|
||||
* **Limited Initial Access**: Limiting initial access to a carefully chosen group to understand and address concerns before wider release.
|
||||
|
||||
### Potential Applications and Impact
|
||||
|
||||
Sora has potential applications in filmmaking, advertising, education, and gaming. It can revolutionize content creation by enabling the creation of realistic and personalized video content and transform educational materials and marketing campaigns.
|
||||
|
||||
However, Sora also has the potential to cause job displacement across various industries, raising concerns about fair compensation for intellectual property rights holders and artists.
|
||||
|
||||
---
|
||||
|
||||
## Key Citations
|
||||
|
||||
- [OpenAI Sora: Text to Video generation - ElevenLabs](https://elevenlabs.io/blog/openai-sora)
|
||||
|
||||
- [Sora (text-to-video model) - Wikipedia](https://en.wikipedia.org/wiki/Sora_(text-to-video_model))
|
||||
|
||||
- [Introducing OpenAI's Sora: Revolutionizing Text-to-Video Conversion](https://pcsocial.medium.com/introducing-openais-sora-revolutionizing-text-to-video-conversion-b99b37a71e55)
|
||||
|
||||
- [OpenAI Sora Is Here! How to Access It and Feature Overview](https://blog.vive.com/us/openai-sora-is-here-how-to-access-it-and-feature-overview/)
|
||||
|
||||
- [What Is OpenAI's Sora? How It Works, Examples, Features](https://www.datacamp.com/blog/openai-announces-sora-text-to-video-generative-ai-is-about-to-go-mainstream)
|
||||
|
||||
- [Six Top Features of Sora, OpenAI's New AI Video Creation Platform](https://www.maginative.com/article/six-top-features-of-sora-openais-new-ai-video-creation-platform/)
|
||||
|
||||
- [The Ultimate Guide to Sora AI + Prompts and Examples - SaaS Genius](https://www.saasgenius.com/blog-business/the-ultimate-guide-to-sora/)
|
||||
|
||||
- [17 Best OpenAI Sora AI Video Examples (2025) - SEO.AI](https://seo.ai/blog/openai-sora-examples)
|
||||
|
||||
- [OpenAI's Sora Video Generator Is Now Available...But Not to All](https://tech.co/news/openai-sora-video-generator-launch)
|
||||
|
||||
- [Generating videos on Sora | OpenAI Help Center](https://help.openai.com/en/articles/9957612-generating-videos-on-sora)
|
||||
|
||||
- [OpenAI Limits Sora Access After Higher-Than-Expected Demand](https://www.pcmag.com/news/openai-releases-sora-video-generator-will-it-simplify-or-destroy-filmmaking)
|
||||
|
||||
- [My OpenAI's Sora video generator review: Is it worth the hype?](https://techpoint.africa/guide/openai-sora-video-generator-review/)
|
||||
|
||||
- [OpenAI disables video gen for certain Sora users as capacity ...](https://techcrunch.com/2025/03/31/openai-disables-video-gen-for-certain-sora-users-as-capacity-challenges-continue/)
|
||||
|
||||
- [Feedback on Sora Text-to-Video Generator. Disappointed - ChatGPT](https://community.openai.com/t/feedback-on-sora-text-to-video-generator-disappointed/1079553)
|
||||
|
||||
- [10 Best Sora AI Prompts For Viral Videos - AI Tools](https://www.godofprompt.ai/blog/10-best-sora-ai-prompts-for-viral-videos?srsltid=AfmBOopcJdIgojxyXbT5TbGxgnr5ijJJAn0dp3wn8net2BmqUzKzCSzS)
|
||||
|
||||
- [Crafting Cinematic Sora Video Prompts: A complete guide · GitHub](https://gist.github.com/ruvnet/e20537eb50866b2d837d4d13b066bd88)
|
||||
|
||||
- [How to Use OpenAI Sora: A Step-by-Step Guide - Alicia Lyttle](https://alicialyttle.com/how-to-use-openai-sora-ai-video-generator/)
|
||||
|
||||
- [Sora: Creating video from text - OpenAI](https://openai.com/index/sora/)
|
||||
|
||||
- [Is OpenAI Sora API Available? And How to Use it? - Apidog](https://apidog.com/blog/openai-sora-api/)
|
||||
|
||||
- [Sora is here - OpenAI](https://openai.com/index/sora-is-here/)
|
||||
|
||||
- [Understanding OpenAI Sora: Features, Uses, and Limitations](https://digitalguider.com/blog/openai-sora/)
|
||||
|
||||
- [Sora's Limitations, Hidden Features and Capabilities (2025)](https://618media.com/en/blog/soras-limitations-and-its-capabilities/)
|
||||
|
||||
- [OpenAI Unveils AI Video Generator Sora, But Limits Human Depictions](https://vocal.media/futurism/open-ai-unveils-ai-video-generator-sora-but-limits-human-depictions)
|
||||
|
||||
- [How to Access Sora in Europe? - Swiftask](https://www.swiftask.ai/blog/comment-acceder-a-sora-en-europe)
|
||||
|
||||
- [How to access Sora AI in UK and EU 2025 - VPNpro](https://vpnpro.com/guides-and-tutorials/how-to-access-sora/)
|
||||
|
||||
- [The Rise of Sora: OpenAI's Frontier in Generative Video Innovation](https://www.launchconsulting.com/posts/the-rise-of-sora-openais-frontier-in-generative-video-innovation)
|
||||
|
||||
- [Meet Sora— OpenAI's Latest Innovation to Bridge the Gap Between Text and Visuals](https://www.practicallogix.com/meet-sora-openais-latest-innovation-to-bridge-the-gap-between-text-and-visuals/)
|
||||
|
||||
- [Do you think the potential use of OpenAI's Sora for creating AI ...](https://www.quora.com/Do-you-think-the-potential-use-of-OpenAIs-Sora-for-creating-AI-deepfakes-are-outweighed-by-the-risks)
|
||||
|
||||
- [What We Know About OpenAI's Sora So Far](https://www.unite.ai/what-we-know-about-openais-sora-so-far/)
|
||||
|
||||
- [What is OpenAI's Sora? and How to Use it? - Great Learning](https://www.mygreatlearning.com/blog/what-is-sora-and-how-to-use-it/)
|
||||
|
||||
- [SORA By Open AI To Kill Off Jobs: Bane or Boon? - Be10X](https://be10x.in/blog/sora-by-open-ai-to-kill-off-jobs-bane-or-boon/)
|
||||
|
||||
- [OpenAI Is Ready for Hollywood to Accept Its Vision](https://www.hollywoodreporter.com/business/business-news/openai-hollywood-sora-1236170402/)
|
||||
54
examples/what_is_agent_to_agent_protocol.md
Normal file
54
examples/what_is_agent_to_agent_protocol.md
Normal file
@ -0,0 +1,54 @@
|
||||
# Google's Agent to Agent Protocol Report
|
||||
|
||||
## Key Points
|
||||
|
||||
- Google's Agent2Agent (A2A) protocol standardizes communication between AI agents, promoting collaboration across diverse systems.
|
||||
- A2A facilitates message exchange for sharing context, instructions, and artifacts between agents.
|
||||
- The protocol complements Anthropic's Model Context Protocol (MCP) by providing a networking layer for agents.
|
||||
- A2A allows agents to negotiate content formats, supporting diverse media types such as iframes and video.
|
||||
- Google intends A2A to be an open, community-driven project to foster innovation and adoption.
|
||||
- Industry experts anticipate A2A will accelerate AI adoption by simplifying integrations and data exchange.
|
||||
|
||||
---
|
||||
|
||||
## Overview
|
||||
|
||||
Google's Agent2Agent (A2A) protocol is designed to establish a standardized method for AI agents to communicate, irrespective of their origin, framework, or location. This initiative seeks to foster seamless collaboration and collective intelligence among AI agents, thereby enhancing the effectiveness of agentic solutions. A2A operates as a networking layer that complements other protocols such as Anthropic's Model Context Protocol (MCP), contributing to a more unified AI agent ecosystem.
|
||||
|
||||
---
|
||||
|
||||
## Detailed Analysis
|
||||
|
||||
### Purpose and Design
|
||||
|
||||
A2A addresses the challenge of integrating disparate AI systems by providing a common language for AI agents. It enables these agents to share context, replies, artifacts, and user instructions, facilitating collaborative problem-solving. The design of A2A supports flexible user experiences by allowing agents to negotiate content formats like iframes, videos, and web forms.
|
||||
|
||||
### Technical Aspects
|
||||
|
||||
The protocol utilizes "parts" within messages, which are fully formed pieces of content with specified content types, enabling negotiation of the correct format needed between agents. A2A builds upon existing standards including HTTP, SSE, and JSON-RPC.
|
||||
|
||||
### Community and Industry Impact
|
||||
|
||||
Google's vision for A2A is to create an open, community-driven project that encourages contributions and updates from the open-source community. Industry experts from companies like Deloitte, Accenture, EPAM, and New Relic believe A2A will accelerate AI adoption by simplifying integrations, facilitating data exchange, and fostering a more unified AI agent ecosystem. LangChain has also expressed interest in collaborating with Google Cloud on this shared protocol.
|
||||
|
||||
### Relationship with MCP
|
||||
|
||||
A2A complements Anthropic's Model Context Protocol (MCP). While A2A provides a networking layer for agents to communicate, MCP functions as a plugin system, granting agents access to tools, context, and data.
|
||||
|
||||
---
|
||||
|
||||
## Key Citations
|
||||
|
||||
- [Announcing the Agent2Agent Protocol (A2A)](https://developers.googleblog.com/en/a2a-a-new-era-of-agent-interoperability/)
|
||||
|
||||
- [Build and manage multi-system agents with Vertex AI - Google Cloud](https://cloud.google.com/blog/products/ai-machine-learning/build-and-manage-multi-system-agents-with-vertex-ai)
|
||||
|
||||
- [Google just Launched Agent2Agent, an Open Protocol for AI agents ...](https://www.maginative.com/article/google-just-launched-agent2agent-an-open-protocol-for-ai-agents-to-work-directly-with-each-other/)
|
||||
|
||||
- [Protocols for Agentic AI: Google's New A2A Joins Viral MCP](https://virtualizationreview.com/articles/2025/04/09/protocols-for-agentic-ai-googles-new-a2a-joins-viral-mcp.aspx)
|
||||
|
||||
- [Google's Agent2Agent interoperability protocol aims to standardize ...](https://venturebeat.com/ai/googles-agent2agent-interoperability-protocol-aims-to-standardize-agentic-communication/)
|
||||
|
||||
- [Meet Google A2A: The Protocol That will Revolutionize Multi-Agent ...](https://medium.com/@the_manoj_desai/meet-google-a2a-the-protocol-that-will-revolutionize-multi-agent-ai-systems-80d55a4583ed)
|
||||
|
||||
- [Google's Agent2Agent Protocol Helps AI Agents Talk to Each Other](https://thenewstack.io/googles-agent2agent-protocol-helps-ai-agents-talk-to-each-other/)
|
||||
@ -113,6 +113,15 @@ def reporter_node(state: State):
|
||||
messages = apply_prompt_template("reporter", state)
|
||||
observations = state.get("observations", [])
|
||||
invoke_messages = messages[:2]
|
||||
|
||||
# Add a reminder about the new report format and citation style
|
||||
invoke_messages.append(
|
||||
HumanMessage(
|
||||
content="IMPORTANT: Structure your report according to the format in the prompt. Remember to include:\n\n1. Key Points - A bulleted list of the most important findings\n2. Overview - A brief introduction to the topic\n3. Detailed Analysis - Organized into logical sections\n4. Survey Note (optional) - For more comprehensive reports\n5. Key Citations - List all references at the end\n\nFor citations, DO NOT include inline citations in the text. Instead, place all citations in the 'Key Citations' section at the end using the format: `- [Source Title](URL)`. Include an empty line between each citation for better readability.",
|
||||
name="system",
|
||||
)
|
||||
)
|
||||
|
||||
for observation in observations:
|
||||
invoke_messages.append(
|
||||
HumanMessage(
|
||||
@ -170,6 +179,15 @@ def _execute_agent_step(
|
||||
]
|
||||
}
|
||||
|
||||
# Add citation reminder for researcher agent
|
||||
if agent_name == "researcher":
|
||||
agent_input["messages"].append(
|
||||
HumanMessage(
|
||||
content="IMPORTANT: DO NOT include inline citations in the text. Instead, track all sources and include a References section at the end using link reference format. Include an empty line between each citation for better readability. Use this format for each reference:\n- [Source Title](URL)\n\n- [Another Source](URL)",
|
||||
name="system",
|
||||
)
|
||||
)
|
||||
|
||||
# Invoke the agent
|
||||
result = agent.invoke(input=agent_input)
|
||||
|
||||
|
||||
@ -15,15 +15,39 @@ You should act as an objective and analytical reporter who:
|
||||
- Never fabricates or assumes information
|
||||
- Clearly distinguishes between facts and analysis
|
||||
|
||||
# Guidelines
|
||||
# Report Structure
|
||||
|
||||
1. Structure your report with:
|
||||
- Executive summary
|
||||
- Key findings
|
||||
- Detailed analysis
|
||||
- Conclusions and recommendations
|
||||
Structure your report in the following format:
|
||||
|
||||
2. Writing style:
|
||||
1. **Key Points**
|
||||
- A bulleted list of the most important findings (4-6 points)
|
||||
- Each point should be concise (1-2 sentences)
|
||||
- Focus on the most significant and actionable information
|
||||
|
||||
2. **Overview**
|
||||
- A brief introduction to the topic (1-2 paragraphs)
|
||||
- Provide context and significance
|
||||
|
||||
3. **Detailed Analysis**
|
||||
- Organize information into logical sections with clear headings
|
||||
- Include relevant subsections as needed
|
||||
- Present information in a structured, easy-to-follow manner
|
||||
- Highlight unexpected or particularly noteworthy details
|
||||
|
||||
4. **Survey Note** (for more comprehensive reports)
|
||||
- A more detailed, academic-style analysis
|
||||
- Include comprehensive sections covering all aspects of the topic
|
||||
- Can include comparative analysis, tables, and detailed feature breakdowns
|
||||
- This section is optional for shorter reports
|
||||
|
||||
5. **Key Citations**
|
||||
- List all references at the end in link reference format
|
||||
- Include an empty line between each citation for better readability
|
||||
- Format: `- [Source Title](URL)`
|
||||
|
||||
# Writing Guidelines
|
||||
|
||||
1. Writing style:
|
||||
- Use professional tone
|
||||
- Be concise and precise
|
||||
- Avoid speculation
|
||||
@ -32,26 +56,28 @@ You should act as an objective and analytical reporter who:
|
||||
- Indicate if data is incomplete or unavailable
|
||||
- Never invent or extrapolate data
|
||||
|
||||
3. Formatting:
|
||||
2. Formatting:
|
||||
- Use proper markdown syntax
|
||||
- Include headers for sections
|
||||
- Use lists and tables when appropriate
|
||||
- Add emphasis for important points
|
||||
- DO NOT include inline citations in the text
|
||||
- Use horizontal rules (---) to separate major sections
|
||||
- Track the sources of information but keep the main text clean and readable
|
||||
|
||||
# Data Integrity
|
||||
|
||||
- Only use information explicitly provided in the input
|
||||
- State "Information not provided" when data is missing
|
||||
- Never create fictional examples or scenarios
|
||||
- If data seems incomplete, ask for clarification
|
||||
- If data seems incomplete, acknowledge the limitations
|
||||
- Do not make assumptions about missing information
|
||||
|
||||
# Notes
|
||||
|
||||
- Start each report with a brief overview
|
||||
- Include relevant data and metrics when available
|
||||
- Conclude with actionable insights
|
||||
- Proofread for clarity and accuracy
|
||||
- Always use the same language as the initial question.
|
||||
- Always use the same language as the initial question
|
||||
- If uncertain about any information, acknowledge the uncertainty
|
||||
- Only include verifiable facts from the provided source material
|
||||
- Place all citations in the "Key Citations" section at the end, not inline in the text
|
||||
- For each citation, use the format: `- [Source Title](URL)`
|
||||
- Include an empty line between each citation for better readability
|
||||
@ -16,16 +16,24 @@ You are dedicated to conducting thorough investigations and providing comprehens
|
||||
4. **Synthesize Information**:
|
||||
- Combine the information gathered from the search results and the crawled content.
|
||||
- Ensure the response is clear, concise, and directly addresses the problem.
|
||||
- Track and attribute all information sources with their respective URLs for proper citation.
|
||||
|
||||
# Output Format
|
||||
|
||||
- Provide a structured response in markdown format.
|
||||
- Include the following sections:
|
||||
- **Problem Statement**: Restate the problem for clarity.
|
||||
- **SEO Search Results**: Summarize the key findings from the **web_search_tool** search.
|
||||
- **Crawled Content**: Summarize the key findings from the **crawl_tool**.
|
||||
- **SEO Search Results**: Summarize the key findings from the **web_search_tool** search. Track the sources of information but DO NOT include inline citations in the text.
|
||||
- **Crawled Content**: Summarize the key findings from the **crawl_tool**. Track the sources of information but DO NOT include inline citations in the text.
|
||||
- **Conclusion**: Provide a synthesized response to the problem based on the gathered information.
|
||||
- **References**: List all sources used with their complete URLs in link reference format at the end of the document. Make sure to include an empty line between each reference for better readability. Use this format for each reference:
|
||||
```
|
||||
- [Source Title](https://example.com/page1)
|
||||
|
||||
- [Source Title](https://example.com/page2)
|
||||
```
|
||||
- Always use the same language as the initial question.
|
||||
- DO NOT include inline citations in the text. Instead, track all sources and list them in the References section at the end using link reference format.
|
||||
|
||||
# Notes
|
||||
|
||||
@ -36,4 +44,6 @@ You are dedicated to conducting thorough investigations and providing comprehens
|
||||
- Do not perform any mathematical calculations.
|
||||
- Do not attempt any file operations.
|
||||
- Only invoke `crawl_tool` when essential information cannot be obtained from search results alone.
|
||||
- Always include source attribution for all information. This is critical for the final report's citations.
|
||||
- When presenting information from multiple sources, clearly indicate which source each piece of information comes from.
|
||||
- Always use the same language as the initial question.
|
||||
|
||||
Loading…
x
Reference in New Issue
Block a user