Lucasmacedo/AI-inovyo-assistende-db
1
0
Fork 0
Code Issues Pull Requests Packages Projects Releases Wiki Activity

Adds final version with new prompt

Browse Source
This commit is contained in:
DNXBrasil
2026-02-11 14:45:00 -03:00
parent c31d089efb
commit 82ac556ecc
9 changed files with 283 additions and 92 deletions
Download Patch File Download Diff File Expand all files Collapse all files

35
code/app/api.py Normal file
View File

@@ -0,0 +1,35 @@
from fastapi import FastAPI
from pydantic import BaseModel
from .backend import BDAgent
app = FastAPI()
@app.get("/")
def health():
return {"status": "ok"}
class QueryRequest(BaseModel):
query: str
history: str = ""
model: str = "anthropic.claude-haiku-4-5-20251001-v1:0"
base: str = "bacio_transacional_loja_app"
class QueryResponse(BaseModel):
response: str
input_tokens: int
output_tokens: int
total_tokens: int
@app.post("/agent", response_model=QueryResponse)
def run_agent(request: QueryRequest):
result = BDAgent.main(request.query, request.history, request.model, request.base)
return QueryResponse(
response=result["response"],
input_tokens=result["input_tokens"],
output_tokens=result["output_tokens"],
total_tokens=result["total_tokens"],
)

188
code/app/backend/BDAgent.py
View File

@@ -4,7 +4,6 @@ LangGraph Agent using AWS Bedrock Cross-Region Inference Profile with Tools
This script demonstrates how to create a LangGraph agent that uses
an AWS Bedrock inference profile with custom tools (add and multiply).
"""
import boto3
from typing import TypedDict, Annotated
from langgraph.graph import StateGraph, END
@@ -18,9 +17,109 @@ from langfuse import Langfuse
from langfuse.langchain import CallbackHandler
from botocore.exceptions import ClientError
import os
from backend.utils import dynamodb_read_table as drt
WORKGROUP = "iceberg-workgroup"
DATABASE = "dnx_warehouse"
TABLE = "poc_dnx_monthly_summary"
REGION = "us-east-1"
# DynamoDB client
dynamodb = boto3.resource("dynamodb", region_name=REGION)
@tool
def get_monthly_report(id: str, variable: str) -> str:
"""
Get a specific variable's data from DynamoDB for a specific id.
Args:
id: The id of the data
variable: The variable/column name to retrieve from the table
Returns:
The content of the specified variable for the given id
"""
print(f"\n🔧 [TOOL CALLED] get_monthly_report for month: {id}, variable: {variable}")
try:
table = dynamodb.Table(TABLE)
response = table.get_item(Key={"id": id})
if "Item" not in response:
return f"No report found for month: {id}"
item = response["Item"]
content = item.get(variable, "")
if not content:
return f"Variable '{variable}' not found for month: {id}"
result = f"<{id}>\n{content}\n</{id}>"
return result
except ClientError as e:
error_message = e.response["Error"]["Message"]
return f"Error fetching report: {error_message}"
@tool
def get_consolidated_keys(id: str) -> str:
"""
Get the list of consolidated keys (variables) available in the table for a specific month.
Args:
id: The id of the data
Returns:
The list of available variables/keys for the specified data
"""
print(f"\n🔧 [TOOL CALLED] get_consolidated_keys for id: {id}")
try:
table = dynamodb.Table(TABLE)
response = table.get_item(Key={"id": id})
if "Item" not in response:
return f"No data found for month: {id}"
item = response["Item"]
chaves_consolidadas = item.get("chaves_consolidadas", "")
if not chaves_consolidadas:
return f"No consolidated keys found for id: {id}"
return chaves_consolidadas
except ClientError as e:
error_message = e.response["Error"]["Message"]
return f"Error fetching consolidated keys: {error_message}"
def get_contexto() -> dict:
"""
Get contexto, filter, and items_disponiveis from DynamoDB where id=DASHBOARD+'_contexto'.
Returns:
Dict with 'contexto', 'filter', and 'items_disponiveis' keys
"""
try:
table = dynamodb.Table(TABLE)
response = table.get_item(Key={"id": DASHBOARD + "_contexto"})
if "Item" not in response:
return {"contexto": "", "filter": "", "items_disponiveis": {}}
item = response["Item"]
return {
"contexto": item.get("contexto", ""),
"filter": item.get("filter_key", ""),
"items_disponiveis": item.get("itens_disponiveis", {}),
}
except ClientError as e:
error_message = e.response["Error"]["Message"]
return {"contexto": f"Error: {error_message}", "filter": "", "items_disponiveis": {}}
def get_secret():
secret_name = "assistente-db-secrets-manager"
@@ -112,7 +211,7 @@ def create_bedrock_llm(model_id: str, region: str = "us-east-1"):
"""
# Determine provider and model_kwargs based on model ID
MODEL_ARNS = {
"anthropic.claude-haiku-4-5-20251001-v1:0": "arn:aws:bedrock:us-east-1:305427701314:inference-profile/global.anthropic.claude-haiku-4-5-20251001-v1:0",
"anthropic.claude-haiku-4-5-20251001-v1:0": "arn:aws:bedrock:us-east-1:305427701314:inference-profile/us.anthropic.claude-haiku-4-5-20251001-v1:0",
"anthropic.claude-sonnet-4-5-20250929-v1:0": "arn:aws:bedrock:us-east-1:305427701314:inference-profile/global.anthropic.claude-sonnet-4-5-20250929-v1:0",
"meta.llama4-maverick-17b-instruct-v1:0": "arn:aws:bedrock:us-east-1:305427701314:inference-profile/us.meta.llama4-maverick-17b-instruct-v1:0",
"meta.llama4-scout-17b-instruct-v1:0": "arn:aws:bedrock:us-east-1:305427701314:inference-profile/us.meta.llama4-scout-17b-instruct-v1:0",
@@ -130,7 +229,7 @@ def create_bedrock_llm(model_id: str, region: str = "us-east-1"):
"amazon.nova-2-lite-v1:0": "amazon"
}
prefix={
"anthropic.claude-haiku-4-5-20251001-v1:0": "global",
"anthropic.claude-haiku-4-5-20251001-v1:0": "us",
"anthropic.claude-sonnet-4-5-20250929-v1:0": "global",
"meta.llama4-maverick-17b-instruct-v1:0": "us",
"meta.llama4-scout-17b-instruct-v1:0": "us",
@@ -147,8 +246,7 @@ def create_bedrock_llm(model_id: str, region: str = "us-east-1"):
)
# Bind tools to the LLM
#tools = [consult_answers,count_table_rows]
tools=[]
tools = [get_monthly_report, get_consolidated_keys]
llm_with_tools = llm.bind_tools(tools)
return llm_with_tools
@@ -160,9 +258,7 @@ def call_model(state: AgentState, llm) -> AgentState:
print(f"[MODEL] Calling Bedrock inference profile...")
messages = state["messages"]
langfuse_handler = CallbackHandler()
config = {"configurable": {"thread_id": "abc123"},"callbacks": [langfuse_handler]}
response = llm.invoke(messages,config=config)
response = llm.invoke(messages)
state["current_step"] = "model_called"
return {"messages": [response]}
@@ -180,6 +276,8 @@ def call_tools(state: AgentState) -> AgentState:
tool_messages = []
tools_map = {
"get_monthly_report": get_monthly_report,
"get_consolidated_keys": get_consolidated_keys
}
# Execute each tool call
@@ -267,8 +365,10 @@ def create_agent(inference_profile_arn: str, region: str = "us-east-1"):
return app
def main(user_query,history,model):
def main(user_query,history,model,base):
"""Main execution function."""
global DASHBOARD
DASHBOARD = base
# Configuration - Update with your actual inference profile ARN
@@ -276,15 +376,41 @@ def main(user_query,history,model):
REGION = "us-east-1"
# System prompt for the agent
SYSTEM_PROMPT=""" You are a analitical agent, with acess to monthly reports about Bacio di latte
contexto_data = get_contexto()
if contexto_data["filter"]=="period":
CONSULT_RULES="""To use the tools you must give the id of the correspondant data, which can be associated to a given month and year in the following format year_month, which:
-Year is the year in 4 digits (2025,2024,2023,2022,2021,...)
-Month is th two digit representation: 01,02,03,04,05,06,07,08,09,10,11,12
The format of the dict is: {id1:year_month1,id2:year_month2...}
Choose the correct id based on the following dict:
"""
elif contexto_data["filter"]=="event":
CONSULT_RULES="""To use the tools you must give the id of the correspondant data, which can be associated to a event, which is in the format "Name - City DD/MM/YYYY", where the last is a date in the format day/month/year. Theformat of elements in dict is {id1:event_description1,id2:event_description2...}"""
else:
CONSULT_RULES="""Wrong filter value, you must terminate the workflow and ask the user to contact the technical team"""
SYSTEM_PROMPT=""" You are a analitical agent in Brazilian Portuguese, with acess to monthly reports about a specific company, specified in the context. You have access to tools that lets you consult present variables in table, you always have access to "context", which keeps inside answers to different questions, that you may consult as you desire.
Do not access other variables besides the ones reported by the tool and "context".
You currently have access to data in a period specified in the context, so only answer questions inside the time window.
<context>
A Bacio di Latte é uma rede de gelaterias artesanais fundada em São Paulo, Brasil, em 2011, pelos irmãos milaneses Edoardo e Luigi Tonolli, que trouxeram a tradição do gelato italiano com ingredientes de alta qualidade, resultando em um sorvete cremoso e fresco, produzido diariamente, sem gordura hidrogenada ou trans, e que se tornou popular não só no Brasil, mas também nos EUA, representando uma experiência autêntica de gelato.
<\context>
<reports>
"""+drt.read_table_as_xml("poc_dnx_monthly_summary","us-east-1")+""""
<\reports>
"""+contexto_data["contexto"]+"""
</context>
"""+CONSULT_RULES+"""
<correlation>
"""+str(contexto_data["items_disponiveis"])+"""
</correlation>
Here is the chat history:"""+history+"""
Aswer the user the best you can with the given information, if you don't know the answer or how to answer say so, only answer from what you know."""
Inside the "NPS" in data is some useful values to calculate the NPS, which includes "distribuicao".
Inside of it are grades and the amount of people who given that grade.
Grades from 0 to 6 are detractors.
Grades from 7 to 8 are neutral.
Grades from 9 to 10 are promoters.
Calculate the percentage of them when prompted about NPS and then calculate the nps using the following formula: NPS = %promoter - %detractor, never use the medium of the notes.
You have access to the tools:
-get_consolidated_keys: Given a id returns the column names inside of a entity of a given table element.
- get_monthly_report: given a id and a variable name, either one listed in the previous tool output or "context", returns its value. Using "context" gives you a summarization of many answers of questions asked to the customers.
Answer, in Brazilian Portuguese, to the user the best you can with the given information, if you don't know the answer or how to answer say so, only answer from what you know.
Always consult the most recent information when a date is not given, like questions "Quanto é meu nps?" """
print("=" * 60)
print("LangGraph Agent with AWS Bedrock Inference Profile + Tools")
@@ -297,11 +423,10 @@ def main(user_query,history,model):
print("\nSystem Prompt: Configured ✓")
print("=" * 60)
# Create the agent
# Create the agent with a unique session_id to group all steps
langfuse_handler = CallbackHandler()
agent = create_agent(INFERENCE_PROFILE_ARN, REGION)
# Example query that requires tools
# Initialize state with system prompt
initial_state = {
"messages": [
@@ -314,8 +439,9 @@ def main(user_query,history,model):
print(f"\nUser Query: {user_query}\n")
print("-" * 60)
# Run the agent
final_state = agent.invoke(initial_state)
# Run the agent with callbacks at graph level
config = {"callbacks": [langfuse_handler], "tags": [DASHBOARD]}
final_state = agent.invoke(initial_state, config=config)
# Display results
print("-" * 60)
@@ -336,7 +462,21 @@ def main(user_query,history,model):
print("\n" + "=" * 60)
print(f"Agent completed successfully. Final step: {final_state['current_step']}")
# Aggregate token usage from all AIMessage objects
total_input_tokens = 0
total_output_tokens = 0
for msg in final_state["messages"]:
if isinstance(msg, AIMessage) and hasattr(msg, 'usage_metadata') and msg.usage_metadata:
total_input_tokens += msg.usage_metadata.get("input_tokens", 0)
total_output_tokens += msg.usage_metadata.get("output_tokens", 0)
langfuse.flush()
return final_state['messages'][-1].content
return {
"response": final_state['messages'][-1].content,
"input_tokens": total_input_tokens,
"output_tokens": total_output_tokens,
"total_tokens": total_input_tokens + total_output_tokens,
}
if __name__=="__main__":
main("oi","ancar_nps_tradicional","","")
main("Liste o nps mês a mês desde maio 2025 até dezembro 2025","","anthropic.claude-sonnet-4-5-20250929-v1:0")

23
code/app/front.py
View File

@@ -2,6 +2,7 @@ import streamlit as st
import time
from backend import BDAgent
import boto3
from boto3.dynamodb.conditions import Key
# Configure the page - MUST BE FIRST
st.set_page_config(
@@ -11,9 +12,19 @@ st.set_page_config(
)
session = boto3.Session()
dynamodb = boto3.resource("dynamodb", region_name="us-east-1")
def list_bases():
table = dynamodb.Table("poc_dnx_monthly_summary")
response = table.query(
IndexName="item_type_index",
KeyConditionExpression=Key("item_type").eq("contexto")
)
return [item["id"].removesuffix("_contexto") for item in response.get("Items", [])]
# Model selection with session state persistence
MODELS = [
"anthropic.claude-haiku-4-5-20251001-v1:0",
"anthropic.claude-sonnet-4-5-20250929-v1:0",
"meta.llama4-maverick-17b-instruct-v1:0",
"meta.llama4-scout-17b-instruct-v1:0",
@@ -27,7 +38,13 @@ selected_value = st.selectbox(
MODELS,
key="selected_model"
)
BASES = list_bases()
BASES.sort()
base = st.selectbox(
"Selecione a base:",
BASES,
key="selected_base"
)
# Initialize chat history in session state
@@ -57,7 +74,8 @@ if prompt := st.chat_input("Type your message here..."):
# Simulate streaming response (replace with actual API call)
full_response = BDAgent.main(prompt,str(st.session_state.messages),selected_value)
result = BDAgent.main(prompt,str(st.session_state.messages),selected_value,base)
full_response = result["response"]
# Simulate typing effect
displayed_response = ""
@@ -67,6 +85,7 @@ if prompt := st.chat_input("Type your message here..."):
time.sleep(0.01)
message_placeholder.markdown(full_response)
st.caption(f"Tokens: {result['input_tokens']} in / {result['output_tokens']} out / {result['total_tokens']} total")
# Add assistant response to chat history
st.session_state.messages.append({"role": "assistant", "content": full_response})

22
code/entrypoint.sh
View File

@@ -1,23 +1,9 @@
#!/bin/bash
# Define o diretório base da aplicação dentro do contêiner
APP_DIR="/app"
# Inicia a API FastAPI em segundo plano
echo "Iniciando API FastAPI na porta 8000..."
uvicorn app.api:app --host 0.0.0.0 --port 8000 &
# Navega para o diretório do backend e inicia a API FastAPI em segundo plano
#echo "Iniciando API FastAPI na porta 8000..."
#python app/backend/main.py &
# Aguarda alguns segundos para garantir que a API tenha tempo de iniciar
# Isso é opcional, mas pode ajudar a evitar problemas de conexão imediata do frontend
#echo "Aguardando a API iniciar..."
#sleep 10 # Ajuste o tempo conforme necessário
# Navega para o diretório do frontend e inicia o aplicativo Streamlit em primeiro plano
# Inicia o aplicativo Streamlit em primeiro plano
echo "Iniciando aplicativo Streamlit na porta 8501..."
# --server.headless=true é importante para rodar Streamlit em ambientes sem GUI (como Docker)
# --server.address=0.0.0.0 permite que o Streamlit seja acessado de fora do contêiner
# --server.enableCORS=false pode ser necessário dependendo da configuração, mas geralmente não para localhost
streamlit run app/front.py --server.port 8501 --server.address 0.0.0.0 --server.headless true
# O comando 'streamlit run' manterá o contêiner em execução.
# Se o Streamlit parar por algum motivo, o script e, consequentemente, o contêiner, terminarão.

2
code/requirements.txt
View File

@@ -4,3 +4,5 @@ langgraph>=0.0.20
langchain>=0.1.0
streamlit
langfuse
fastapi
uvicorn

0
code/app/backend/utils/dynamodb_read_table.py → code/utils/dynamodb_read_table.py
View File

12
infra/ecs_alb/Pulumi.Inovyo.yaml
View File

@@ -21,12 +21,12 @@ config:
- 3.14.44.224/32
ecs_subnet_ids:
- subnet-0f50f25a2fbb054d4
- subnet-0014ea77951bbeb6c
- subnet-043a427630309c2f4
app-ecs:ecs:
- task_name: assisnte-analitico-db-dev
ecr_repo_name: assistente-analitico-db-dev
ecr_image_tag: latest
ecr_image_digest: sha256:7a2aede33d8dd34822b73291d64e1ccce26980fc531290c54d992bcb7dee26fa
ecr_image_digest: sha256:c9cd83a7caf51e7eee471d47649bbe0157d4c2703f365010f12997e63d941d3d
cpu: 256
memory: 512
desired_count: 1
@@ -36,11 +36,15 @@ config:
min_capacity: 1
max_capacity: 3
target_value: 60.0
lb_config:
name: listener
lb_configs:
- name: streamlit
listener_port: 8501
target_port: 8501
container_port: 8501
- name: api
listener_port: 8000
target_port: 8000
container_port: 8000
env_variables:
LANGFUSE_HOST: http://172.31.252.176:3000
# SECRET_NAME: dev/ai-pge-doc-classification

80
infra/ecs_alb/ecs.py
View File

@@ -6,33 +6,42 @@ import ecr
import json
def deploy_app(config_ecs_app, app_ecs_cluster, alb_security_group, app_load_balancer_arn):
lb_config = config_ecs_app["lb_config"]
target_group = aws.lb.TargetGroup(f"app-target-group-{lb_config['listener_port']}",
port=lb_config["target_port"],
protocol="HTTP",
vpc_id=config.network["vpc_id"],
target_type="ip",
health_check=aws.lb.TargetGroupHealthCheckArgs(
path="/", # TODO if it doesn't work, can use /docs for fastapi
protocol="HTTP",
port="traffic-port",
healthy_threshold=2,
unhealthy_threshold=2,
timeout=5,
interval=30,
matcher="200-499",
),
)
aws.lb.Listener(f"app-listener-{lb_config['listener_port']}",
load_balancer_arn=app_load_balancer_arn,
port=lb_config["listener_port"],
protocol="HTTP",
default_actions=[aws.lb.ListenerDefaultActionArgs(
type="forward",
target_group_arn=target_group.arn,
)],
)
# target_groups.append(target_group)
lb_configs = config_ecs_app["lb_configs"]
target_groups = []
load_balancers = []
for lb_config in lb_configs:
tg = aws.lb.TargetGroup(f"app-target-group-{lb_config['listener_port']}",
port=lb_config["target_port"],
protocol="HTTP",
vpc_id=config.network["vpc_id"],
target_type="ip",
health_check=aws.lb.TargetGroupHealthCheckArgs(
path="/",
protocol="HTTP",
port="traffic-port",
healthy_threshold=2,
unhealthy_threshold=2,
timeout=5,
interval=30,
matcher="200-499",
),
)
aws.lb.Listener(f"app-listener-{lb_config['listener_port']}",
load_balancer_arn=app_load_balancer_arn,
port=lb_config["listener_port"],
protocol="HTTP",
default_actions=[aws.lb.ListenerDefaultActionArgs(
type="forward",
target_group_arn=tg.arn,
)],
)
target_groups.append(tg)
load_balancers.append(aws.ecs.ServiceLoadBalancerArgs(
target_group_arn=tg.arn,
container_name=f"{config.project_name}-{config_ecs_app['task_name']}-{config.environment}-service",
container_port=lb_config["target_port"],
))
# Build and Push ECR
# ecr_repos = ecr.create_ecr_repo(config_ecs_app['ecr_repo_name'])
@@ -79,12 +88,11 @@ def deploy_app(config_ecs_app, app_ecs_cluster, alb_security_group, app_load_bal
"cpu": 0,
"portMappings": [
{
"name": "api",
"containerPort": lb_config["container_port"],
"hostPort": lb_config["target_port"],
"name": lb_cfg["name"],
"containerPort": lb_cfg["container_port"],
"hostPort": lb_cfg["target_port"],
"protocol": "tcp",
}
# } for lb_config_item in lb_configs
} for lb_cfg in lb_configs
],
"essential": True,
"logConfiguration": {
@@ -135,12 +143,6 @@ def deploy_app(config_ecs_app, app_ecs_cluster, alb_security_group, app_load_bal
)
# ECS Service Setup
load_balancer = aws.ecs.ServiceLoadBalancerArgs(
target_group_arn=target_group.arn,
container_name=f"{config.project_name}-{config_ecs_app['task_name']}-{config.environment}-service",
container_port=lb_config["target_port"],
)
app_service = aws.ecs.Service(f"{config.project_name}-{config_ecs_app['task_name']}-service",
cluster=app_ecs_cluster.arn,
task_definition=app_task_definition.arn,
@@ -151,7 +153,7 @@ def deploy_app(config_ecs_app, app_ecs_cluster, alb_security_group, app_load_bal
security_groups=[app_ecs_security_group.id],
subnets=config.network["ecs_subnet_ids"],
),
load_balancers=[load_balancer],
load_balancers=load_balancers,
scheduling_strategy="REPLICA",
service_connect_configuration=aws.ecs.ServiceServiceConnectConfigurationArgs(
enabled=True,

5
infra/ecs_alb/iam.py
View File

@@ -186,7 +186,10 @@ def create_task_role():
"dynamodb:Query",
"dynamodb:DescribeTable"
],
"Resource": "arn:aws:dynamodb:us-east-1:305427701314:table/poc_dnx_monthly_summary"
"Resource": [
"arn:aws:dynamodb:us-east-1:305427701314:table/poc_dnx_monthly_summary",
"arn:aws:dynamodb:us-east-1:305427701314:table/poc_dnx_monthly_summary/index/*"
]
},
],
}),