|
|
""" |
|
|
Agent used to determine which specialized agents should be used to address knowledge gaps. |
|
|
|
|
|
The Agent takes as input a string in the following format: |
|
|
=========================================================== |
|
|
ORIGINAL QUERY: <original user query> |
|
|
|
|
|
KNOWLEDGE GAP TO ADDRESS: <knowledge gap that needs to be addressed> |
|
|
=========================================================== |
|
|
|
|
|
The Agent then: |
|
|
1. Analyzes the knowledge gap to determine which agents are best suited to address it |
|
|
2. Returns an AgentSelectionPlan object containing a list of AgentTask objects |
|
|
|
|
|
The available agents are: |
|
|
- WebSearchAgent: General web search for broad topics |
|
|
- SiteCrawlerAgent: Crawl the pages of a specific website to retrieve information about it |
|
|
""" |
|
|
|
|
|
|
|
|
try: |
|
|
from ..utils.llm_client import model_supports_structured_output, qianwen_plus_model |
|
|
from ..utils.baseclass import ResearchAgent |
|
|
from ..utils.parse_output import create_type_parser |
|
|
except ImportError: |
|
|
|
|
|
from utils.llm_client import model_supports_structured_output, qianwen_plus_model |
|
|
from utils.baseclass import ResearchAgent |
|
|
from utils.parse_output import create_type_parser |
|
|
from datetime import datetime |
|
|
from typing import List, Optional |
|
|
|
|
|
from pydantic import BaseModel, Field |
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
class AgentTask(BaseModel): |
|
|
"""A task for a specific agent to address knowledge gaps""" |
|
|
|
|
|
gap: Optional[str] = Field( |
|
|
description="The knowledge gap being addressed", default=None |
|
|
) |
|
|
agent: str = Field(description="The name of the agent to use") |
|
|
query: str = Field(description="The specific query for the agent,should be str") |
|
|
entity_website: Optional[str] = Field( |
|
|
description="The citation of the article,include author,publication year,journal name, if known", |
|
|
default=None, |
|
|
) |
|
|
|
|
|
|
|
|
class AgentSelectionPlan(BaseModel): |
|
|
"""Plan for which agents to use for knowledge gaps""" |
|
|
|
|
|
tasks: List[AgentTask] = Field( |
|
|
description="List of agent tasks to address knowledge gaps" |
|
|
) |
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
INSTRUCTIONS = f""" |
|
|
You are an Tool Selector responsible for determining which specialized agents should address a knowledge gap in a research project. |
|
|
Today's date is {datetime.now().strftime("%Y-%m-%d")}. |
|
|
|
|
|
You will be given: |
|
|
1. The original user query |
|
|
2. A knowledge gap identified in the research |
|
|
3. A full history of the tasks, actions, findings and thoughts you've made up until this point in the research process |
|
|
|
|
|
Your task is to decide: |
|
|
1. Which specialized agents are best suited to address the gap |
|
|
2. What specific queries should be given to the agents (keep most 100 words) |
|
|
|
|
|
Available specialized agents: |
|
|
- ArticleSearchAgent: search the web for information relevant to the query - provide a query with most 100 words as , - use this most 3 times if you want to find out something about a the information of query,the refs are in the output of this agent |
|
|
|
|
|
Guidelines: |
|
|
- Aim to call at most 2 agents at a time in your final output |
|
|
- You can list the PubMedSearchAgent at most 1 times with different queries if needed to cover the full scope of the knowledge gap |
|
|
- Be specific and concise (most 100 words) with the agent queries - they should target exactly what information is needed |
|
|
- If you know the citation of the article of an entity being researched, always include it in the query |
|
|
- If a gap doesn't clearly match any agent's capability, default to generate query and search PubMedSearchAgent |
|
|
- Use the history of actions / tool calls as a guide - try not to repeat yourself if an approach didn't work previously |
|
|
- For the citation of the article: |
|
|
1. Use ONLY information that is explicitly provided in the articles |
|
|
2. DO NOT invent or fabricate any information, dates, journal names, or other details |
|
|
3. For missing information, use "N/A" or omit the field entirely, but NEVER invent data |
|
|
4. Use this format: Author(s), (Year). Title. Journal, Volume(Issue), Pages. |
|
|
5. If any piece of information is missing, simply exclude it rather than making it up |
|
|
|
|
|
For example, if author, year and title are available but not journal details: |
|
|
- Smith J, Johnson K. (2020). Advances in gene therapy for cancer treatment. |
|
|
|
|
|
|
|
|
Only output JSON. Follow the JSON schema below. Do not output anything else. I will be parsing this with Pydantic so output valid JSON only: |
|
|
{AgentSelectionPlan.model_json_schema()} |
|
|
""" |
|
|
|
|
|
selected_model = qianwen_plus_model |
|
|
|
|
|
tool_selector_agent = ResearchAgent( |
|
|
name="ToolSelectorAgent", |
|
|
instructions=INSTRUCTIONS, |
|
|
model=selected_model, |
|
|
output_type=AgentSelectionPlan |
|
|
if model_supports_structured_output(selected_model) |
|
|
else None, |
|
|
output_parser=( |
|
|
create_type_parser(AgentSelectionPlan) |
|
|
if not model_supports_structured_output(selected_model) |
|
|
else None |
|
|
), |
|
|
) |
|
|
|
