Writing IDE MVP

A Docs-as-Code AI writing workspace with a React/Tiptap frontend and a FastAPI backend fully migrated onto the fastapi-fullstack / full-stack-ai-agent-template structure.

Current repository structure:

• frontend: active React + Tiptap workspace
• desktop: planned Electron shell
• backend: active Python backend generated from the template and adapted to this product
• nginx / docker-compose*.yml / Makefile: template-aligned operational assets

The current product surface implements:

• three-panel writing workspace
• right-side AI chat panel with enter-to-send, multi-turn history, and streaming assistant output
• selection-aware AI rewrite flow with status streaming, diff preview, and accept/reject
• mock document store with multiple Markdown drafts
• manual version snapshots, history diff, and restore
• Markdown export preview from the current editor state
• FastAPI rewrite backend at /api/v1/ai/rewrite/*
• workspace-aware agent chat over /api/v1/ws/agent with Read, Write, Glob, Grep, and WebSearch tools

The next active product track continues hardening the right sidebar agentic chat surface with:

• enter-to-send chat input
• multi-turn conversation state
• streaming assistant message updates
• no-selection task execution over the current workspace
• agent tools such as Read, Write, Glob, Grep, and WebSearch

Development

npm install
npm run dev

Create a local .env first:

cp .env.example .env

npm run dev runs the frontend only.

Required environment variables:

• OPENAI_API_KEY
• OPENAI_MODEL (default: gpt-4o-mini)
• OPENAI_BASE_URL (default: https://api.openai.com/v1)

Optional local backend variables:

• POSTGRES_HOST, POSTGRES_PORT, POSTGRES_USER, POSTGRES_PASSWORD, POSTGRES_DB
• SECRET_KEY
• CORS_ORIGINS
• LOG_LEVEL, LOG_DIR, LOG_MAX_BYTES, LOG_BACKUP_COUNT

Fresh Machine Setup

For a clean machine, bring the project up in this order:

1. Install Node.js, uv, and Docker.
2. Install dependencies with npm install and make install.
3. Create a local env file with cp .env.example .env.
4. Set OPENAI_API_KEY, OPENAI_MODEL, and OPENAI_BASE_URL in .env.
5. Start PostgreSQL with make docker-db.
6. Run migrations with make db-upgrade.
7. Start the app with make run.

Important local database notes:

• The project Docker PostgreSQL is exposed on host port 5433, not 5432.
• .env.example is aligned to 5433 for local make run.
• If you already run PostgreSQL directly on your machine, either keep this project's Docker DB on 5433 or change .env to point at your existing instance.
• backend/.env.example remains 5432 because the backend container talks to the db container over the internal Docker network.

Recommended smoke checks after startup:

• make db-current
• Open http://localhost:5173
• Send one message in the right-side AI Chat
• If something fails, inspect logs/latest/app.log and logs/latest/error.log

Build

npm run build

Full Stack

make install
make run

Useful commands:

• npm run dev
• make run-backend
• make test
• make routes
• uv run --project backend docgent_backend --version

Backend logs are written to logs/ at the repository root by default. Each backend start creates a fresh logs/runs/<timestamp-pid>/ directory with app.log and error.log, and logs/latest points to the newest run so the current session can be inspected directly from the workspace.

Documentation

• Chinese README: docs/README_CN.md
• Docs hub: docs/product-specs/index.md
• Design docs: docs/design-docs/index.md
• Active execution plans: docs/PLANS.md