Documentation Index
Fetch the complete documentation index at: https://mintlify.com/vrashmanyu605-eng/Langchain_Interview_Multi_Agents_Flow/llms.txt
Use this file to discover all available pages before exploring further.
The email agent is the final specialist in the pipeline. It takes the completed candidate profile, all three interview question sets, and the hiring evaluation, then drafts three separate emails — one for each interviewer — so every interviewer enters the session with a tailored brief.
Source code
from llm import llm
def email_agent(state):
candidate_profile = state["candidate_profile"]
hr_questions = state["hr_questions"]
technical_questions = state["technical_questions"]
ceo_questions = state["ceo_questions"]
evaluation = state["evaluation"]
response = llm.invoke(
f"""
You are an Email Communication Agent.
Generate professional interview emails.
Candidate:
{candidate_profile}
HR Questions:
{hr_questions}
Technical Questions:
{technical_questions}
CEO Questions:
{ceo_questions}
Evaluation:
{evaluation}
Generate:
- HR email
- technical interviewer email
- CEO interviewer email
Include:
- candidate summary
- interview focus
- attached materials summary
Return structured JSON only.
"""
)
return {
"email_content": response.content
}
candidate_profile, hr_questions, technical_questions, ceo_questions, evaluation
Provides: email_content
The agent reads five state fields that are all produced by earlier agents. The supervisor ensures all five are present before routing here.
Output structure
The agent returns email_content as a JSON string containing three email objects:
{
"hr_email": {
"subject": "Interview Brief: Jane Smith — HR Round",
"body": "Hi Sarah,\n\nPlease find below the interview brief for Jane Smith...",
"candidate_summary": "Senior Backend Engineer with 5 years of Python and API development experience.",
"interview_focus": ["Communication style", "Team conflict handling", "Ownership mindset"],
"materials_summary": "Resume and HR question bank (15 questions) attached."
},
"technical_email": {
"subject": "Interview Brief: Jane Smith — Technical Round",
"body": "Hi Alex,\n\nHere is the technical interview brief for Jane Smith...",
"candidate_summary": "Strong Python backend engineer with system design experience.",
"interview_focus": ["System design", "API scalability", "Debugging approach"],
"materials_summary": "Resume and 20 technical questions with expected answers attached."
},
"ceo_email": {
"subject": "Interview Brief: Jane Smith — CEO Round",
"body": "Hi CEO,\n\nPlease find the culture-fit brief for Jane Smith...",
"candidate_summary": "Senior engineer showing strong ownership and strategic thinking.",
"interview_focus": ["Long-term vision", "Business impact mindset", "Leadership philosophy"],
"materials_summary": "Resume, evaluation summary, and CEO question bank (10 questions) attached."
}
}
Role in the pipeline
The email agent is always the last agent called before the workflow ends. Once email_content is present in state, the supervisor routes to "finished" and the pipeline stops.
| Email | Recipient | Focus |
|---|
hr_email | HR interviewer | Behavioral and culture questions |
technical_email | Technical interviewer | Architecture, coding, and debugging |
ceo_email | CEO/exec interviewer | Leadership, strategy, and vision |
Accessing email output
In main.py, the email agent’s output is printed as:
elif node == "email":
print(state.get("email_content"))
final_state after the loop completes:
import json
email_data = json.loads(final_state.get("email_content", "{}"))
hr_email = email_data.get("hr_email", {})
technical_email = email_data.get("technical_email", {})
ceo_email = email_data.get("ceo_email", {})