Sprint 6: SQL Analytics Agent

Goal

Add structured financial analytics using SEC Company Facts stored in PostgreSQL.

Why This Sprint Matters

Not every financial question should be answered through RAG. Revenue, net income, assets, liabilities, cash, operating cash flow, and share counts are structured facts that are better queried through safe SQL templates.

What Was Built

Financial facts table
SEC Company Facts client and AAPL sample fallback
POST /api/ingest/company-facts
POST /api/sql/analyze
Safe metric templates with no raw SQL input
LangGraph SQL route
Frontend SQL Analytics controls
sql-smoke evaluation suite

Architecture / Workflow

mermaid

flowchart LR
    SECFacts[SEC Company Facts] --> Ingest[Company Facts Ingestion]
    Ingest --> Facts[(PostgreSQL Financial Facts)]
    Request[SQL Analyze Request] --> Templates[Safe Query Templates]
    Templates --> Facts
    Facts --> Summary[Trend Summary]
    Summary --> Response[SQL Analytics Response]

Key Files And APIs

backend/app/services/company_facts_client.py
backend/app/services/sql_analytics_service.py
POST /api/ingest/company-facts
POST /api/sql/analyze

Validation Commands

powershell

Invoke-RestMethod -Method Post http://localhost:8000/api/sql/analyze `
  -ContentType "application/json" `
  -Body '{"ticker":"AAPL","metric":"revenue","period":"annual","limit":5}'

Demo Talking Points

Explain why the project avoids LLM-generated SQL in the MVP. Safe templates reduce injection risk and make evaluation deterministic.

What Changed From Previous Sprint

Sprint 5 routed questions across agents. Sprint 6 adds a new structured-data route.

Sprint 6: SQL Analytics Agent ​

Goal ​

Why This Sprint Matters ​

What Was Built ​

Architecture / Workflow ​

Key Files And APIs ​

Validation Commands ​

Demo Talking Points ​

What Changed From Previous Sprint ​