💡 No API key yet? Try Fluxion AI — one entry point to access and manage leading global AI models. OpenAI-compatible, works with Vela out of the box, $3 free credit on signup. 👉 Claim now

---

Vela is an open-source, privacy-first, local-first AI writing IDE purpose-built for novel writing, web fiction, and creative writing. It deeply integrates LLM-powered workflows (outline generation, chapter drafting, intelligent rewriting, automated review) with a local RAG knowledge base, giving authors an IDE-level immersive creative experience — all running on your own machine with your own API keys (BYOK).

---

Screenshots

Immersive writing workspace with side-by-side AI panel, IDE-grade window management

End-to-end AI novel writing pipeline: from worldbuilding to chapter generation

More Screenshots

---

Key Features

Vela is not just another chat-based text editor — it is a production-grade novel writing engine that deeply integrates LLM capabilities, long-context retrieval (RAG), and automated pipelines for fiction authoring.

AI-Powered Novel Writing Pipeline

Capability	Description
Worldbuilding	Custom global world settings, core plot axes, character profiles (with cross-chapter dynamic state tracking)
Auto Outline	One-click AI generation of "structural skeleton → chapter outlines → scene/emotion/rhythm requirements", supports Three-Act, Hero's Journey and other narrative structures
Chapter Drafting	Single-chapter streaming generation, accurately responding to prior context and preset outlines, can be stopped at any time
AI Rewrite	Supports paragraph-level or full-chapter rewriting while maintaining character and plot consistency
Refine	AI automatically detects grammar errors, typos, and logic gaps, outputting polished suggestions
Review	AI evaluates chapters from a reader/editor perspective, identifying pacing, character arc, and foreshadowing issues
Triple Post-Process Pipeline	Rewrite → Refine → Review three-stage chain ensuring high-quality chapter output

Million-Word Local RAG Knowledge Base

Capability	Description
Bulk Import	One-click import of millions of words of reference novels, world settings documents, character setting collections
Vector Semantic Search	Automatically recalls the most relevant setting chunks based on current chapter semantics — say goodbye to character/setting inconsistencies
Pure Local Storage	Built-in SQLite + lightweight vector engine, all data stored locally, works offline

Extensible Architecture

Capability	Description
BYOK (Bring Your Own Key)	Natively compatible with OpenAI, DeepSeek, Gemini, Claude, Ollama (local offline), Zhipu GLM, etc. Smart routing: use DeepSeek for outlines, Claude for polishing, local models for privacy review
MCP Protocol	Native Model Context Protocol integration, attach custom tool servers to extend AI capabilities
Usage Analytics	Built-in statistics panel for LLM calls, token consumption, and cost trends

IDE-Grade Productivity UI

Capability	Description
Resizable Panels	File tree + editor + AI panel + bottom terminal, flexible combination like VSCode/JetBrains
Dark Theme	Optimized dark mode with custom floating title bar and status bar micro-interactions
Keyboard Shortcuts	Global shortcuts: Cmd+N new, Cmd+O open, Cmd+=/- zoom
Cross-Platform	macOS (dmg) / Windows (nsis) / Linux (AppImage)

---

Installation

Direct Download

Go to Releases to download the latest version for your OS:

• macOS: .dmg installer
• Windows: .exe installer (NSIS)

Build from Source

# Requirements: Node.js >= 18, npm >= 9

# 1. Clone the project
git clone https://github.com/Izya12/vela.git
cd vela

# 2. Install dependencies
npm install

# 3. Start dev server (with hot reload)
npm run dev

# 4. Build for distribution
npm run build

Note: You need build tools for native SQLite modules (macOS: Xcode Command Line Tools, Windows: Visual Studio Build Tools).

---

Model Configuration

Vela supports multiple mainstream LLM providers. Quick setup:

1. Open the app → click Settings in the bottom-left
2. Go to Model Configuration
3. Click Add Model:
   • Select provider: OpenAI / DeepSeek / Gemini / Ollama / Zhipu / Custom
   • Fill in API Key and Base URL (if using a proxy)
   • Assign recommended models for different tasks (writing / polishing / Embedding retrieval)
4. Start writing!

Supported LLM Providers:

OpenAI · DeepSeek · Google Gemini · Anthropic Claude · Ollama (Local) · Zhipu GLM · MiniMax · SiliconFlow · Fluxion AI (Recommended) · Any OpenAI-compatible API

💡 Recommended: if you don't have an official key or want to cut costs, use Fluxion AI (OpenAI-compatible). In step 3 above choose "Custom / OpenAI-compatible" and fill in your Fluxion AI Base URL + API Key. Get $3 free credit on signup.

---

🌟 Recommended: Fluxion AI — One Entry Point for Leading Global AI Models

Fluxion AI serves indie developers, tech teams and enterprises with a unified API to access and manage leading global AI models. Multi-route dynamic scheduling improves availability, with transparent model performance, latency and billing. When using Fable 5.1, Fluxion AI can save up to ~90% vs. official Claude API pricing.

• ✅ Unified OpenAI-Compatible API, works with Vela out of the box
• 🔀 Multi-route dynamic scheduling for higher availability
• 📊 Transparent model performance, latency and costs
• 🎁 Get $3 API credit on signup

👉 Exclusive link: Sign up and claim $3 credit

---

Sponsor

Vela's open-source edition is driven by passion in spare time. If this tool has improved your writing efficiency or you see its commercial potential, feel free to sponsor! Your support is my biggest motivation to keep iterating.

---

Tech Stack

Layer	Technology
UI Framework	React 19 + TypeScript + Zustand
Styling	Tailwind CSS v4 + Radix UI + Lucide Icons
Desktop	Electron 41 + Vite 8
Local Storage	better-sqlite3 (relational) + Lightweight vector engine (RAG)
IPC	Type-safe IPC channels
AI Integration	OpenAI-compatible + Gemini Protocol + MCP

---

Contributing

We welcome community contributions, including but not limited to:

• Bug fixes
• New AI provider adapters
• UI/UX improvements
• Internationalization (i18n) translations
• Documentation improvements

For major feature refactors, please discuss with the author first in Issues to avoid direction conflicts.

---

License

This project is licensed under GPL-3.0 License. You are free to run, study, share, and modify the code, but new software based on this modified distribution must also be open-sourced under GPL-3.0.

For closed-source commercial licensing, please contact the author via email.

---