add recruiter example

examples/recruiter/langgraph.json

@@ -0,0 +1,7 @@
+{
+  "dependencies": ["."],
+  "graphs": {
+    "research": "./recruiter_agent.py:create_agent"
+  },
+  "env": ".env"
+}

examples/recruiter/recruiter_agent.py

@@ -0,0 +1,73 @@
+import os
+
+from exa_py import Exa
+
+from deepagents import create_deep_agent, SubAgent
+
+exa = Exa(api_key = os.environ['EXA_API_KEY'])
+
+
+
+
+
+
+# Search tool to use to do research
+def linkedin_search(
+    query: str,
+    num_results: int = 5,
+):
+    """Run a linkedin search"""
+    return exa.search_and_contents(
+        query,
+        text=True,
+        num_results=num_results,
+        type="auto",
+        category="linkedin profile"
+    )
+
+
+sub_research_prompt = """You are a dedicated researcher. Your job is to source candidates for the role described.
+
+Write down any new candidates in candidates.json
+
+only include candidates if they look good
+
+Respond to the user saying how many candidates you wrote down."""
+
+research_sub_agent = {
+    "name": "research-agent",
+    "description": "Used to kick off in depth searches. Give it a brief to research.",
+    "prompt": sub_research_prompt,
+    "tools": ["linkedin_search"]
+}
+
+# Prompt prefix to steer the agent to be an expert researcher
+research_instructions = """You are an expert sourcer
+
+Use the research agent to run specific searches. It will write its results to candidates.json"""
+
+# Create the agent
+agent = create_deep_agent(
+    [linkedin_search],
+    research_instructions,
+    subagents=[research_sub_agent],
+).with_config({"recursion_limit": 1000})
+
+from pydantic import BaseModel
+
+class Config(BaseModel):
+    instructions: str = research_instructions
+    subagents: list[SubAgent] = [research_sub_agent]
+
+from langchain_core.runnables import RunnableConfig
+
+def create_agent(config: RunnableConfig):
+    config = config.get('configurable', {})
+    config_fields = {k: v for k,v in config.items() if k in ['instructions', 'subagents']}
+    config = Config(**config_fields)
+    return create_deep_agent(
+        [linkedin_search],
+        config.instructions,
+        subagents=config.subagents,
+        config_schema=Config,
+    ).with_config({"recursion_limit": 1000})

examples/recruiter/requirements.txt

@@ -0,0 +1,3 @@
+deepagents
+langgraph-cli[inmem]
+exa-py

src/deepagents/sub_agent.py

@@ -2,7 +2,7 @@
 from deepagents.state import DeepAgentState
 from langgraph.prebuilt import create_react_agent
 from langchain_core.tools import BaseTool
-from typing import TypedDict
+from typing_extensions import TypedDict
 from langchain_core.tools import tool, InjectedToolCallId
 from langchain_core.messages import ToolMessage
 from typing import Annotated, NotRequired