Deploy and Host Piney on Railway

Piney (Little Brother Xu) is a dedicated workstation designed for managing SillyTavern character cards, world info, and chat logs. It offers deep analysis of V2/V3 card specs, AI-assisted content generation, token usage analytics, interactive chat replay, image gallery management with ComfyUI and NovelAI metadata parsing, and automated data backups.

About Hosting Piney

Hosting Piney on Railway provides a containerized environment using its official prebuilt Docker image. Railway runs the application process, handles inbound HTTP routing, and manages TLS termination automatically.

Because Piney processes and stores character card specifications, world info databases, custom assets, and system configurations locally, persistent storage is required to prevent data loss across container redeployments. Railway provisions a persistent volume mounted directly to the application container, ensuring data durability. Networking is established by configuring an HTTP proxy to map external web traffic to Piney's internal service port. The application runs as a standalone server instance without needing external database services like PostgreSQL or Redis.

Common Use Cases

Managing, creating, and analyzing SillyTavern character cards (V2/V3 specifications) and World Info entries with AI assistance.
Importing, reviewing, and replaying SillyTavern chat histories to track token consumption and turn counts.
Organizing character illustrations in a smart gallery with automatic ComfyUI/NovelAI metadata extraction and AI tagging.

Dependencies for Piney Hosting

Railway Volume (for persistent application data storage)

Deployment Dependencies

Implementation Details

Docker

Piney is deployed using the prebuilt Docker image:

ghcr.io/andclear/piney:latest

No local container build is necessary.

Public Networking

Configure public networking under Railway Service Settings:

Setting	Value
Proxy Type	HTTP Proxy
Target Port	`9696`

Environment Variables

Variable	Required	Description
`HOST`	Yes	Network host binding interface (`0.0.0.0`).
`PORT`	Yes	Internal application listening port (`9696`).
`RUN_MODE`	Yes	Mode of operation for the application server (`server`).

Raw Editor format:

HOST=0.0.0.0
PORT=9696
RUN_MODE=server

Persistent Storage

Piney requires persistent volume storage to preserve character cards, galleries, and configurations.

Open the Piney service in Railway.
Go to Settings → Volumes.
Click Add Volume.
Set the Mount Path:

/app/data

This volume stores all persistent user data, exported backup files, and local application states.

Build & Start

No custom build or start commands are required. Railway uses the entrypoint and command configurations specified inside the prebuilt Docker container image.

Accessing the Application

In your Railway service dashboard, navigate to Settings → Networking.
Click Generate Domain to create a public *.up.railway.app URL.
Open the generated domain in your web browser to access the Piney web interface.

Application Features

📊 Kanban Page

Light and dark modes: Supports switching between light and dark modes.
Data statistics: Statistics on the total capacity of character cards, short plays, images, and data in the current database.
Today's fortune: You can draw three cards each day (3 random cards each time), and the lucky card for today will be replaced after confirmation.

📚 Character Library

Management functions: As the entry point, cards are sorted in reverse order by the last edited time by default.
Privacy protection: Supports enabling/disabling cover blur display ("little eye" function).
Right-click menu: Right-click (long press on mobile devices) the cover to perform operations such as multi-select deletion (soft deletion), moving to category, and export.
Categories and Tags: Supports category management and tag filtering functions.

📝 Character Card Details Page

AI Assistance: Supports AI-powered generation of character card descriptions and tags (API Key configuration required).
Cover Management: Supports changing the cover (WebP, JPG, PNG).
Token statistics: Supports token statistics for reference only.
Notes system: Supports adding notes.

🩺 Dr. Pi (Beta)

The AI can diagnose character cards, world books, and other items, and provide suggestions for improvement.

Intelligent detection: AI will intelligently filter relevant world book entries for reading to avoid token explosion.
Detection scope: Setting diagnosis, plot logic, character consistency, worldview logic, OOC warning, etc.
Test report: Automatically stored in the database (For reference only, entertainment function).

⚙️ Settings Management

Simplified configuration: Displays basic configuration items, reducing card writing complexity. Supports AI optimization and AI translation of all text areas.
Full-screen editing: All text areas support full-screen editing.
Creator Information: The creator information of "Imported" cards cannot be modified; cards created by the user can be claimed in "Version History".
Opening Remarks Preview: Supports direct preview of opening remarks styles (including multiple opening remarks).
AI Character Generation: Only supports AI generation of character cards created by the user (including task settings, etc.).

🌍 Character Card with Embedded World Book

Simplified management: Retains the most frequently used functions and supports AI optimization/translation of entries.
Mutual guidance function:
Supports importing entries from the global world book into the current character.
Supports exporting the current character's World Book entries to the global World Book.

🧩 Character Card Embedded Regular Expression

Simplified functionality: "Advanced Mode" supports settings consistent with SillyTavern.
Code Editor: The "Full Screen Editing" function for regular expression entries features a code editor style that supports real-time preview rendering effects and code formatting.

💬 Quick Character Card Recharge

Supports importing/exporting quick reply files.

🕰 Version History & Claim

Save mechanism: Similar to the save/load (SL) method in games, after creating a version history, modifications can be rolled back.
Claiming Your Work: At the bottom of the version history page, you can claim your own character card, unlocking AI-generated descriptions/worldbooks/opening lines and the ability to change your author's name.

💬 Chat History

Format support: Supports importing TXT and JSONL formats. JSONL will be converted to TXT by default.
Follow the Wind Mode: Only when this mode is enabled will the original JSONL format be imported and preserved.
Regular expression rendering: Supports configuring preset regular expression files to render and preview chat history.
Reading progress: Automatically saves the last reading position.

🎭 Mini Theater

Collection Management: Save your favorite short clips, with support for color tags and automatic categorization.
Format compatibility: Batch import/export is compatible with the "Echo Theater" format.

🌐 The Global World Book

Supports importing and editing global world books, with the same functionality as the world book embedded in the character card.

🖼️ Image Gallery

Image management: Supports categorization, favorites (pinned to the top), and color filtering.
Search: Supports searching by name, tags, categories, and positive suggestions.
Metadata recognition: Automatically recognizes images (original images) generated by NovelAI/ComfyUI, tags AI-generated information, and retains prompt words.

📥 Import

Only character cards (JSON/PNG) and the global world book are supported for import. PNGs must contain character information.

✨ Pipi Beautification Workbench

(API configuration required)

AI Enhancement: Based on the original text and input content, AI outputs regular expression scripts, style codes, and world book entries.
Real-time rendering: Supports real-time preview of generated styles and "click to edit" (AI-based targeted modifications).
One-click insertion: The generated resources can be inserted into the current character card with one click.

🗑️ Recycling Station

Only character cards are moved to the recycle bin after deletion. The recycle bin retains characters indefinitely by default and requires manual cleanup.

📦 Data Backup

Supports one-click full export (.piney format) and recovery. (Note: The restore operation is a full overwrite, and all original data will be lost).

⚙️ System Settings

API settings: Supports OpenAI format API, multi-channel configuration, and one-click detection.
Global prompt: After setting, it will be injected at the very beginning of all AI calls (except for front-end beautification).

Why Deploy Piney on Railway?

Railway is a singular platform to deploy your infrastructure stack. Railway will host your infrastructure so you don't have to deal with configuration, while allowing you to vertically and horizontally scale it.

By deploying Piney on Railway, you are one step closer to supporting a complete full-stack application with minimal burden. Host your servers, databases, AI agents, and more on Railway.