Summary

StoryForge is a privacy-first, local-first, browser-based AI writing studio designed for long-form fiction authors. It addresses the common issue of opaque AI prompt systems by providing full visibility and control over AI outputs. The tool empowers writers with advanced features for managing complex narratives, worldbuilding, and character consistency, aiming to amplify creative control rather than replace it.

How It Works

This project employs a local-first architecture, storing manuscripts in the browser's IndexedDB by default for enhanced privacy. It integrates with over a dozen AI providers (including OpenAI, Claude, Gemini) via OpenAI-compatible APIs, requiring users to supply their own API keys. A core differentiator is its "no black box" prompt system, where every prompt is visible, editable, and savable, allowing authors granular control. The system supports series-scale work through multi-world settings, a three-layer memory system, and a codex for worldbuilding and consistency checking.

Quick Start & Requirements

To begin, clone the repository, navigate to the directory, and run npm install followed by npm run dev. This will start a local development server accessible at http://localhost:1111/storyforge/. The primary prerequisite is Node.js LTS. Windows users should download the source ZIP and use npm for installation and execution.

Highlighted Details

• Prompt Engineering Control: Features a fully visible, editable, and savable prompt system with system prompts and user templates. Offers three levels of adjustment: parameter sliders, text overrides, and temporary edits.
• Extensive AI Vendor Support: Integrates with numerous AI providers (OpenAI, Claude, Gemini, Kimi, etc.) and supports local Ollama, requiring users to manage their own API keys.
• Advanced Narrative Management: Implements a three-layer memory system (Working, Episodic, Semantic), a codex for worldbuilding, consistency checking, and reverse-inference from text fragments.
• Historical Fiction Focus: Includes specialized tools for historical research, timeline management, detailed analysis of historical documents, and context-aware prompts for historical accuracy.
• Workflow Engine: Enables chaining AI generation steps into automated workflows (e.g., story core to chapter generation) with visual progress tracking and automatic data saving.

Maintenance & Community

The project is maintained by yuanbw2025. Community interaction is encouraged via a QQ group (1082374587) and GitHub Discussions for sharing prompt templates and workflows. Contributions are welcomed following the guidelines in CONTRIBUTING.md. Development progress is tracked through distinct "Phases."

Licensing & Compatibility

The project's license policy is explicitly stated as "not yet finalized." Users are advised to complete LICENSE, EULA, and privacy policy documentation before commercial release or open-source distribution. Compatibility is primarily for individual users leveraging local browser storage and their own AI service accounts.

Limitations & Caveats

The most significant limitation is the undefined licensing status, which presents a substantial barrier for commercial adoption or open-source distribution. Users must provide and manage their own API keys for AI services, and the local-first nature means cloud backup and AI features are dependent on user configuration and external service availability.