Multi_agent / app.py
spandana30's picture
Update app.py
a7524b6 verified
raw
history blame
7.57 kB
import streamlit as st
import os
import time
import base64
from typing import Dict, List, TypedDict
from langgraph.graph import StateGraph, END
from huggingface_hub import InferenceClient
client = InferenceClient(
model="mistralai/Mistral-7B-Instruct-v0.2",
token=st.secrets["HF_TOKEN"]
)
class AgentState(TypedDict):
messages: List[Dict[str, str]]
design_specs: str
html: str
css: str
feedback: str
iteration: int
done: bool
timings: Dict[str, float]
PRODUCT_MANAGER_PROMPT = """You're a product manager. Given the user request:
{user_request}
Break it down into clear features and priorities."""
PROJECT_MANAGER_PROMPT = """You're a project manager. Based on these features:
{features}
Draft a quick development plan with key tasks and timeline."""
ARCHITECT_PROMPT = """You're a software architect. Create design specs for:
{user_request}
Include:
1. Color palette (primary, secondary, accent)
2. Font choices
3. Layout structure
4. Component styles
Don't write code - just design guidance."""
ENGINEER_PROMPT = """Create a complete HTML page with embedded CSS for:
{design_specs}
Requirements:
1. Full HTML document with <!DOCTYPE>
2. CSS inside <style> tags in head
3. Mobile-responsive
4. Semantic HTML
5. Ready-to-use (will work when saved as .html)
Output JUST the complete HTML file content:"""
QA_PROMPT = """Review this website:
{html}
Check for:
1. Visual quality
2. Responsiveness
3. Functionality
Reply "APPROVED" if perfect, or suggest improvements."""
def call_model(prompt: str, max_retries=3) -> str:
for attempt in range(max_retries):
try:
return client.text_generation(
prompt,
max_new_tokens=3000,
temperature=0.3,
return_full_text=False
)
except Exception as e:
st.error(f"Model call failed (attempt {attempt+1}): {str(e)}")
st.write("\n\n**Full Error:**", e)
st.stop() # Force exit to show error
return "<html><body><h1>Error generating UI</h1></body></html>"
def time_agent(agent_func, state: AgentState, label: str):
start = time.time()
result = agent_func(state)
duration = time.time() - start
result["timings"] = state["timings"]
result["timings"][label] = duration
return result
def product_manager_agent(state: AgentState):
features = call_model(PRODUCT_MANAGER_PROMPT.format(user_request=state["messages"][-1]["content"]))
return {"messages": state["messages"] + [{"role": "product_manager", "content": features}]}
def project_manager_agent(state: AgentState):
features_msg = next((m["content"] for m in state["messages"] if m["role"] == "product_manager"), "")
plan = call_model(PROJECT_MANAGER_PROMPT.format(features=features_msg))
return {"messages": state["messages"] + [{"role": "project_manager", "content": plan}]}
def software_architect_agent(state: AgentState):
specs = call_model(ARCHITECT_PROMPT.format(user_request=state["messages"][-1]["content"]))
return {"design_specs": specs, "messages": state["messages"] + [{"role": "software_architect", "content": specs}]}
def engineer_agent(state: AgentState):
html = call_model(ENGINEER_PROMPT.format(design_specs=state["design_specs"]))
if not html.strip().startswith("<!DOCTYPE"):
html = f"""<!DOCTYPE html>
<html><head><meta charset='UTF-8'><meta name='viewport' content='width=device-width, initial-scale=1.0'>
<title>Generated UI</title></head><body>{html}</body></html>"""
return {"html": html, "messages": state["messages"] + [{"role": "software_engineer", "content": html}]}
def qa_agent(state: AgentState, max_iter: int):
feedback = call_model(QA_PROMPT.format(html=state["html"]))
done = "APPROVED" in feedback or state["iteration"] >= max_iter
return {"feedback": feedback, "done": done, "iteration": state["iteration"] + 1,
"messages": state["messages"] + [{"role": "qa", "content": feedback}]}
def generate_ui(user_request: str, max_iter: int):
state = {"messages": [{"role": "user", "content": user_request}], "design_specs": "", "html": "",
"css": "", "feedback": "", "iteration": 0, "done": False, "timings": {}}
workflow = StateGraph(AgentState)
workflow.add_node("product_manager", lambda s: time_agent(product_manager_agent, s, "product_manager"))
workflow.add_node("project_manager", lambda s: time_agent(project_manager_agent, s, "project_manager"))
workflow.add_node("software_architect", lambda s: time_agent(software_architect_agent, s, "software_architect"))
workflow.add_node("software_engineer", lambda s: time_agent(engineer_agent, s, "software_engineer"))
workflow.add_node("qa", lambda s: time_agent(lambda x: qa_agent(x, max_iter), s, "qa"))
workflow.add_edge("product_manager", "project_manager")
workflow.add_edge("project_manager", "software_architect")
workflow.add_edge("software_architect", "software_engineer")
workflow.add_edge("software_engineer", "qa")
workflow.add_conditional_edges("qa", lambda s: END if s["done"] else "software_engineer")
workflow.set_entry_point("product_manager")
app = workflow.compile()
total_start = time.time()
final_state = app.invoke(state)
return final_state["html"], final_state, time.time() - total_start
def main():
st.set_page_config(page_title="Multi-Agent Collaboration", layout="wide")
st.title("🀝 Multi-Agent Collaboration")
with st.sidebar:
max_iter = st.slider("Max QA Iterations", 1, 5, 2)
st.write("\nπŸ” HF_TOKEN found:", st.secrets.get("HF_TOKEN", "❌ Missing!"))
prompt = st.text_area("πŸ“ Describe the UI you want:", "A coffee shop landing page with hero, menu, and contact form.", height=150)
if st.button("πŸš€ Generate UI"):
with st.spinner("Agents working..."):
html, final_state, total_time = generate_ui(prompt, max_iter)
st.success("βœ… UI Generated Successfully!")
st.components.v1.html(html, height=600, scrolling=True)
st.subheader("πŸ“… Download HTML")
b64 = base64.b64encode(html.encode()).decode()
st.markdown(f'<a href="data:file/html;base64,{b64}" download="ui.html">Download HTML</a>', unsafe_allow_html=True)
st.subheader("🧐 Agent Communication Log")
history_text = ""
for msg in final_state["messages"]:
role = msg["role"].replace("_", " ").title()
content = msg["content"]
history_text += f"---\n{role}:\n{content}\n\n"
st.text_area("Agent Dialogue", value=history_text, height=300)
b64_hist = base64.b64encode(history_text.encode()).decode()
st.markdown(
f'<a href="data:file/txt;base64,{b64_hist}" download="agent_communication.txt" '
'style="padding: 0.4em 1em; background: #4CAF50; color: white; border-radius: 0.3em; text-decoration: none;">'
'πŸ“… Download Communication Log</a>',
unsafe_allow_html=True
)
st.subheader("πŸ“Š Performance")
st.write(f"⏱️ Total Time: {total_time:.2f} seconds")
st.write(f"πŸ” Iterations: {final_state['iteration']}")
for stage in ["product_manager", "project_manager", "software_architect", "software_engineer", "qa"]:
st.write(f"πŸ€– {stage.title().replace('_', ' ')} Time: {final_state['timings'].get(stage, 0):.2f}s")
if __name__ == "__main__":
main()