InvokeAI Reports

A web-based analytics dashboard that reads your InvokeAI installation's SQLite database and presents usage statistics through interactive charts and tables. Understand your image generation patterns — model preferences, prompt habits, parameter trends, and more.

Features

• Model Analytics — Top models, family distribution, model leaderboard with usage stats
• Prompt Analysis — Token frequency, word clouds, prompt length distribution
• Trend Tracking — Generation volume over time, parameter evolution, activity heatmaps
• Generation Stats — Resolution distribution, scheduler usage, LoRA statistics, error rates
• Unused Models — Installed models that were never used in any generation (filterable by main/LoRA, with size totals to help reclaim disk space)
• Multi-User — Auto-detects users from InvokeAI, filterable per-user or aggregate
• Date Filtering — Custom date ranges with preset shortcuts (7d, 30d, 90d, all-time)

Architecture

InvokeAI DB (invokeai.db) → Importer (read-only) → App DB (reports.db) → FastAPI → React UI

The importer reads InvokeAI's database in read-only mode, parses JSON metadata, and flattens it into the app's own reports.db. All analytics queries run against the app database, never the source.

Quick Start

Prerequisites

• Python 3.10+
• uv (recommended) or pip
• Node.js 18+ (only needed if building the frontend from source)
• An InvokeAI installation with a databases/invokeai.db file

Installation

git clone https://github.com/cochbild/invoke-ai-reports.git
cd invoke-ai-reports

# Set up Python backend (uv handles venv + locked deps)
uv sync

# Build the frontend
cd frontend
npm install
npm run build
cd ..

Prefer plain pip? python -m venv .venv && .venv/bin/pip install -e . works too — pyproject.toml is the source of truth.

Running

All platforms:

uv run uvicorn backend.app.main:app --port 9876

Open http://localhost:9876 in your browser. On first run, enter your InvokeAI installation path (e.g., G:\InvokeUi, C:\Users\you\InvokeAI, or /home/user/invokeai).

Without uv, activate your venv first (source .venv/bin/activate on Linux/Mac, .venv\Scripts\activate on Windows) and drop the uv run prefix.

Open http://localhost:9876. When prompted for the InvokeAI path:

• Native Linux: Enter the path directly (e.g., /home/user/invokeai)
• WSL accessing a Windows drive: Use the WSL mount path (e.g., /mnt/g/InvokeUi) or the Windows path (G:\InvokeUi) — both are accepted

Path Flexibility

The setup page accepts any of these path formats:

Input	What it finds
G:\InvokeUi	G:\InvokeUi\databases\invokeai.db
G:\InvokeUi\databases	G:\InvokeUi\databases\invokeai.db
G:\InvokeUi\databases\invokeai.db	Exact file path
/mnt/g/InvokeUi	/mnt/g/InvokeUi/databases/invokeai.db (WSL)

Configuration

Environment variables (prefix INVOKE_REPORTS_):

Variable	Default	Description
INVOKE_REPORTS_PORT	9876	Server port
INVOKE_REPORTS_HOST	127.0.0.1	Server host
INVOKE_REPORTS_APP_DB_PATH	reports.db	Path to app database
INVOKE_REPORTS_DEV_CORS	false	Allow Vite dev server (:5173) origin — set when running npm run dev

Development

Backend

# Sync deps including dev tooling (pytest, httpx, pip-audit)
uv sync --dev

# Run backend tests
uv run pytest -v

# Run backend with hot reload
uv run uvicorn backend.app.main:app --reload --port 9876

# Audit dependencies for known CVEs
uv run pip-audit

Frontend

cd frontend

# Install dependencies
npm install

# Run dev server (proxies /api to backend on port 9876)
npm run dev

# Run frontend tests
npx vitest run

# Production build
npm run build

During development, run the backend (uvicorn ... --port 9876) and frontend dev server (npm run dev) simultaneously. The Vite dev server proxies /api requests to the backend.

API Endpoints

All stats endpoints accept optional user_id, start_date, and end_date query parameters.

Method	Path	Description
POST	/api/sync	Trigger data import
GET	/api/sync/status	Last sync info
GET	/api/users	List users
GET	/api/stats/overview	KPI summary
GET	/api/stats/models/top	Top models by usage
GET	/api/stats/models/least	Least used models
GET	/api/stats/models/unused	Installed models never used (lifetime); ?type=main|lora and ?search=…
GET	/api/stats/models/family-distribution	Model family breakdown
GET	/api/stats/models/leaderboard	Full model stats table
GET	/api/stats/prompts/top-tokens	Most frequent prompt tokens
GET	/api/stats/prompts/length-distribution	Prompt length histogram
GET	/api/stats/trends/volume	Generation count over time
GET	/api/stats/trends/heatmap	Activity by day/hour
GET	/api/stats/trends/parameters	Avg steps/CFG over time
GET	/api/stats/generation/resolutions	Resolution frequency
GET	/api/stats/generation/schedulers	Scheduler distribution
GET	/api/stats/generation/steps	Steps histogram
GET	/api/stats/generation/cfg	CFG scale histogram
GET	/api/stats/generation/loras	LoRA usage stats
GET	/api/stats/generation/errors	Failure rates and error types
POST	/api/validate-path	Check InvokeAI DB path
GET	/api/settings	Current config
PUT	/api/settings	Update config
POST	/api/settings/clear	Reset to first-run state

Tech Stack

Layer	Technology
Backend	Python, FastAPI, SQLAlchemy, SQLite
Frontend	React, TypeScript, Vite, Chakra UI, Recharts
Testing	pytest, Vitest

License

MIT