Skip to content

Latest commit

 

History

History
118 lines (83 loc) · 2.64 KB

File metadata and controls

118 lines (83 loc) · 2.64 KB

Pebble

Status Frontend Backend AI Service Queue Data

Pebble turns technical manuals into structured, editable checklists with citation-aware AI support.

Reason through manuals, do not just keyword search them.

Pebble Overview

Overview

Pebble helps teams transform dense maintenance and operations documentation into practical, auditable checklist workflows.

Documentation

Architecture

Stack

  • Frontend: React + Vite (apps/web)
  • API: Express + TypeScript (services/api)
  • AI Service: FastAPI (services/ai)
  • Data: MongoDB + Redis/BullMQ

Request flow:

Web -> /api/* (Express) -> /v1/* (FastAPI)

Async checklist jobs:

Express route -> BullMQ queue -> API worker -> AI pipeline -> MongoDB

Features

  • Auth (register/login/me)
  • Manual upload and management
  • Checklist generation jobs with status tracking
  • Checklist detail page with item updates
  • PDF export for generated checklists
  • Per-manual persisted chat history

Repository Structure

  • apps/web: React frontend
  • services/api: Express API + worker
  • services/ai: FastAPI AI pipeline service
  • docs: project documentation
  • packages/shared-types: shared DTO package scaffold

Quick Start

Requirements:

  • Node.js 20+
  • npm 10+
  • Python 3.11+
  • MongoDB
  • Redis
  • OpenAI-compatible local model endpoint (for example Docker Model Runner)

Install:

cd apps/web && npm install
cd ../../services/api && npm install
cd ../ai && pip install -r requirements.txt

Run (4 terminals):

# 1) AI service
cd services/ai
uvicorn app.main:app --reload --port 8001
# 2) API service
cd services/api
npm run dev
# 3) Worker service
cd services/api
npm run worker:dev
# 4) Web app
cd apps/web
npm run dev

Health checks:

  • GET http://localhost:4000/health
  • GET http://localhost:4000/health/deps
  • GET http://localhost:8001/health