Disclosure Pilot AI

Disclosure Pilot AI is a full-stack desktop application that demonstrates using Large Language Models to automate the drafting, compilation, and iXBRL compliance tagging of SEC quarterly Form 10-Q and yearly Form 10-K financial reports.

For comprehensive documentation and visual guides, see [docs/Disclosure-Pilot-AI.pdf].

Key Features

• Automated Filing Assembly: Assembles modular document components (Cover Page, Financial Statements, MD&A, Notes) into master .docx reports and EDGAR-compliant HTML files.

• AI Financial Agents: Employs local ChatOllama agents via LangGraph to perform financial calculations, enforce SEC compliance, and draft narrative sections.

• iXBRL Integration: Converts HTML filings into Inline XBRL (iXBRL) format validated against US-GAAP taxonomies using Arelle.

Tech Stack

Category	Technologies / Tools
Frontend	React 18, Vite, Axios, Recharts
Backend	FastAPI, Uvicorn, SqlAlchemy, Python 3.13
Database	PostgreSQL 18 (read-only financial ledger)
AI & Agent Orchestration	ChatOllama, LangGraph, LangChain
Document Processing	python-docx, docxcompose, custom iXBRL postprocessor (xbrl_tagger.py)
iXBRL Viewer	Arelle validator/viewer
CLI	Windows PowerShell scripts

For full installation instructions, see [docs/installation/Start-Here-Setup.docx].

Prerequisites

• OS: Windows 11
• Python: Version 3.13 and added to system PATH
• Node.js: Installed and nodejs added to system PATH
• PostgreSQL: Running local instance and bin added to system PATH
• Ollama: Running service with local LLM models pre-loaded and ollama.exe added to system PATH
• Arelle GUI: Installed and arelleGUI.exe added to system PATH

---

Installation Overview

1. Clone the Repository Open Windows PowerShell (standard user, non-admin) and run:

git clone https://github.com/p-dirac/disclosure-pilot-ai
cd disclosure-pilot-ai

2. Populate Database Open pgAdmin, launch the Query Tool for your target database, and execute the following SQL scripts to set up the tables and schema:

• db-applogins.sql
• db-bookkeeper.sql

3. Extract AppIO Files Unpack and install the required AppIO files from AppIO-latest.zip into the working directory.

4. Configure Environment Variables Create the root .env file using the provided template:

Copy-Item .env.example .env

Edit .env to set your PostgreSQL connection details, local Ollama model parameters, and JWT secret key.

5. Unblock Setup Scripts Allow execution of local PowerShell scripts:

Unblock-File .\unlock.ps1

6. Install & Compile Run the automated installer to set up the Python virtual environment (backend\.venv), install backend dependencies, compile the React frontend, and stage build files into backend\static:

.\install.ps1

Running the Project

1. Verify Connections Run the preflight connectivity checker to confirm that PostgreSQL, Ollama, and Arelle endpoints are reachable:

.\check-connections.ps1

2. Launch Application Start the FastAPI web server. This will automatically open the application in your default browser at http://localhost:8000:

.\run.ps1

3. Log In Use the default test credentials to authenticate:

   • Email: zx@qaz.com
   • Password: qwerty

4. Menu Structure Follow this workflow: Prep → Report → EDGAR.

   • Home
   • 10-Q: Prep, Report, EDGAR
   • 10-K: Prep, Report, EDGAR
   • Help: User Guide, About

Directory Architecture

• Application Repository:

disclosure-pilot-ai/
├── backend/
│   ├── app/
│   │   ├── agents/    # AI agents 
│   │   ├── api/       # FastAPI route endpoints
│   │   ├── core/      # App configurations
│   │   ├── models/    # User auth database
│   │   ├── schemas/   # Input/output validation
│   │   └── services/  # Core business logic layer
│   └── tests/      # Backend unit tests
├── docs/       # Project description & setup
└── frontend/
	├── dist/     # Index files
	└── src/
		├── api/         # Axios clients & HTTP 
		├── components/  # UI menus, tables, dialogs
		├── hooks/       # User auth context
		├── pages/       # Top-level views
		├── styles/      # Global CSS module
		└── tests/       # Frontend unit tests

• Document Workspace: (stores report components and compiled final outputs)

AppIO/
├── reports/
├── sec10k/
├── sec10q-q1/
├── sec10q-q2/
├── sec10q-q3/
└── user_input/