ToolExperimental

cidadao.ai-backend

Name: cidadao.ai-backend
Rating: 3.0 (35 reviews)
Author: anderson-ufrj

by anderson-ufrj

Multi-agent backend that converts Brazil public data into investigative insights

Python

Updated Sep 18, 2026

Stars

Forks

Commits/Month

View on GitHub

Summary

Transforms raw Brazilian public transparency data into structured, investigator-friendly insights using a multi-agent backend. Agents perform data ingestion, NLP extraction, and investigative synthesis, coordinated by Python services and exposed via a FastAPI backend. Distinctive focus on civic tech workflows: pipeline components tuned for open-data from the Portal da Transparência and workflows for turning records into human-readable investigations.

Avoid common pitfalls

Learn what failures to watch for

Why It Matters

As civic applications automate data triage and hypothesis generation, knowing how agents agent delegation and produce evidence matters for trust. This project makes agent delegation explicit for public-data investigations, so outputs can be traced back to specific ingestion and analysis steps. That traceability is valuable for building agent track records and for later Agent-to-Agent Protocol (A2A) evaluation or reputation scoring in civic scenarios.

Best For

Civic-tech teams building investigative pipelines from government open-data who need an agentized ingestion-to-insight backend. For more robust task routing, consider the Dynamic Task Routing Pattern.

Applications

Extracting and normalizing fiscal records from the Portal da Transparência for downstream analysis
Running coordinated agent pipelines to summarize, link, and surface anomalies in public procurement
Tracing which agent modules produced specific evidence to support auditability of automated investigations