Previo

Collaborative Markdown + LaTeX writing platform

---

☕ Support the Project

If you find Previo useful and want to support its development:

👉 https://buymeacoffee.com/r4m0

Your support helps maintain and improve the platform ❤️

---

Highlights

• Authentication: Email/password + Google/GitHub OAuth
• Projects: Markdown/LaTeX editing, import .md/.tex, search, delete
• Collaboration: Invite by email, accept/deny workflow, shared access
• Versioning: Snapshot history, diff view, save comments, revert
• Rendering: Markdown preview/export + LaTeX preview/export (PDF)
• Workspace: Per-project file/image uploads accessible from LaTeX
• Workspace References: use /projectId/path/to/file.ext inside Markdown/LaTeX (resolved in preview/export)
• AI Co-Writer: local Ollama-powered draft generation + in-editor assistant
• AI Conversations: create/switch/delete conversations with persisted history
• CLI: Cross-platform previo start <port> / previo stop

Screenshots

Auth

Projects

Profile

Collaboration

Version History

Workspace

AI Co-Writer

LaTeX Features

Project Structure

previo/
├── backend/
│   ├── db/postgres_service.py
│   ├── latex/
│   ├── markdown/
│   └── requirements.txt
├── ui/
│   ├── app/
│   ├── components/
│   ├── lib/
│   ├── types/
│   └── package.json
├── screenshots/
├── bin/previo.js
└── package.json

Requirements

• Node.js 18+
• npm
• Python 3.10+
• PostgreSQL 14+ (or Docker Compose)
• LaTeX compiler (pdflatex)
• Docker + Docker Compose (recommended for full stack)
• Ollama (required for AI features)

Quick Start (Recommended)

1. Clone and enter project:

git clone https://github.com/R4M-0/previo.git
cd previo

2. Configure .env at repo root (copy from .env.example):

POSTGRES_DB=previo
POSTGRES_USER=previo
POSTGRES_PASSWORD=previo
PREVIO_DATABASE_URL=postgresql://previo:previo@postgres:5432/previo
NEXT_PUBLIC_APP_URL=http://localhost:3000
GOOGLE_CLIENT_ID=
GOOGLE_CLIENT_SECRET=
GITHUB_CLIENT_ID=
GITHUB_CLIENT_SECRET=
PREVIO_OLLAMA_BASE_URL=http://host.docker.internal:11434
PREVIO_OLLAMA_MODEL=llama3.1:8b
PREVIO_OLLAMA_TIMEOUT_MS=300000

3. Install and start Ollama on your host machine:

# install (Linux/macOS)
curl -fsSL https://ollama.com/install.sh | sh

# pull a model used by Previo (example default)
ollama pull llama3.1:8b

On Linux, make Ollama reachable from Docker containers:

sudo systemctl edit ollama

Add:

[Service]
Environment="OLLAMA_HOST=0.0.0.0:11434"

Then restart:

sudo systemctl daemon-reload
sudo systemctl restart ollama

Quick check:

curl http://localhost:11434/api/tags

4. Install the CLI once:

npm link

5. Start app:

previo start 3000

6. Stop app:

previo stop

Open http://localhost:3000.

Local Development (Without CLI)

UI dependencies

cd ui
npm install

Python dependencies

python3 -m venv .venv
source .venv/bin/activate
pip install -r backend/requirements.txt

Run dev server

cd ui
npm run dev

Open http://localhost:3000.

AI Co-Writer Setup Details

Previo uses Ollama via environment variables only:

• PREVIO_OLLAMA_BASE_URL: Ollama endpoint reachable by the UI container
• PREVIO_OLLAMA_MODEL: model name to use for generate/chat
• PREVIO_OLLAMA_TIMEOUT_MS: timeout per request in milliseconds

Optional tuning variables:

• PREVIO_OLLAMA_DRAFT_MAX_TOKENS
• PREVIO_OLLAMA_CHAT_MAX_TOKENS
• PREVIO_OLLAMA_NUM_CTX
• PREVIO_OLLAMA_TEMPERATURE
• PREVIO_OLLAMA_CHAT_MAX_HISTORY

By default, Previo does not hard-cap generation/context for chat and draft quality (unless you set the optional vars above).

Common values:

• Dockerized app + Ollama on host:
  • PREVIO_OLLAMA_BASE_URL=http://host.docker.internal:11434
• Non-dockerized local app + Ollama on same machine:
  • PREVIO_OLLAMA_BASE_URL=http://localhost:11434

If AI requests time out often:

• keep your model warmed up (ollama run <model> once),
• increase PREVIO_OLLAMA_TIMEOUT_MS (e.g. 420000),
• use a smaller/faster model.

Docker Notes

• docker compose up uses docker-compose.override.yml (dev mode with bind mounts)
• docker compose -f docker-compose.yml up --build uses production image
• Workspace files are persisted in workspaces_data and available under PREVIO_WORKSPACES_DIR
• For AI features, Ollama must run on host and be reachable from container (PREVIO_OLLAMA_BASE_URL)

OAuth Callback URLs

• Google: http://localhost:3000/api/auth/oauth/google/callback
• GitHub: http://localhost:3000/api/auth/oauth/github/callback

Use a consistent host (localhost recommended).

Main Features

Auth

• Signup/login with email/password
• Google OAuth
• GitHub OAuth
• Session cookie auth

Projects

• Create Markdown or LaTeX projects
• Import existing .md / .tex
• Search by title/content/format
• Delete owned projects

Collaboration

• Invite collaborators by email
• Accept or deny invitations
• Shared editing access for accepted members
• Remote save sync in open editors

Versioning

• Auto snapshot on save
• Optional save comments
• Diff viewer for each version
• Revert to any previous snapshot

Rendering

• Markdown preview + HTML export
• LaTeX preview + PDF export

Workspace

• Per-project file storage
• Upload/list/open/delete files
• Copy and use shorthand references like /projectId/assets/diagram.png
• Shorthand references are auto-resolved in Markdown/LaTeX preview and export

AI Co-Writer

• AI draft generation during project creation (description-based)
• In-editor chat assistant for Markdown and LaTeX
• Replace suggestions with Approve/Modify workflow
• Multi-conversation management (new/switch/delete)
• Conversation history persisted in PostgreSQL

License

This project is licensed under the MIT License. See the LICENSE file for details.

Additional Docs

• UI: ui/README.md
• Backend: backend/README.md