·
4 commits
to main
since this release
PaperChat v1.1.0-beta - Release Notes
Minor Update Release
This release introduces significant improvements to PaperChat's existing ecosystem and user experience, configuration system and AI provider integration. The plugin now offers a more intuitive command structure, comprehensive configuration options and support for Hack Club's AI models, alongside existing Google and OpenAI providers.
What does Beta 1.1.0 Offer
Core Features
- Simplified Command Structure: Commands now target the sender. Cross-player AI communication is no longer available due to possible abuse
- Chat History Management: New
/paperchat historyand/paperchat clearcommands - Tab Autocompletion: Smart autocompletion for history and clear subcommands
- HackClub AI Provider: New free provider supporting ai.hackclub.com with multiple models (available only for Hack Club's members)
- Enhanced Configuration: Support for both config.yml and environment variables
AI Provider System
- Multi-Provider Support: Google Gemini, OpenAI GPT, and HackClub AI models
- Flexible Authentication: Providers can be keyless if the underlying service doesn't require API keys
- Automatic Response Sanitization: JSON-Sanitizer integration for robust response parsing
- Think Tag Removal: Automatic filtering of OpenAI reasoning content from responses
- Model Configuration: Improved model selection and validation per provider
Configuration & Securityrity
- Dual Configuration System:
config.ymlfile configuration alongside existing environment variables - Configuration Priority: Environment variables > t;
config.yml> default valuesues - API Key Security: Optional API keys based on provider requirements
- Input Validation: Character limits and response sanitization
- System Prompt Customization: Improved default AI behavior for better user experience
Development Features
- JSON Sanitization Library: Added Google's JSON-Sanitizer for malformed response handling
- Enhanced Provider Architecture: Cleaner separation between provider logic and response processing
- Improved Error Handling: Better configuration validation and user-friendly error messages
- Build System Updates: Updated Gradle dependencies and shadow plugin configuration
Breaking Changes
Command Structure Changes
- Previous:
/paperchat;message> - Current:
/paperchat(AI responds directly to you)ge>
Configuration Changes
- HackClub provider uses empty model field (server-side model selection)
- Google and OpenAI providers now require explicit model configuration
- API keys are now optional and provider-dependent
Technical Requirements
Server Requirements
- PaperMC 1.21 or higher
- Java 21 or higher
- Minimum 1GB RAM (2GB+ recommended)
API Requirements
- Google AI Studio API key (for Google provider)
- OpenAI API key (for OpenAI provider)
- No API key (for HackClub provider - free for Hack Club members)
- Internet connectivity for API requests
Installation & Upgraderade
Fresh Installation
- Download the JAR from releases
- Place in your server's
plugins/directory - Launch the server to generate default config
- Configure via
config.ymlor environment variables (check README.md for more details) - Restart your server
Upgrading from v1.0.0-beta
- Replace the old JAR with the new version
- Update configuration to specify explicit models for Google/OpenAI providers
- Optionally migrate to config.yml configuration
- Restart your server
Usage
Send an AI-powered message:
/paperchat How do I build a redstone contraption?
/paperchat What are the best mining strategies?
Manage your chat history:
/paperchat history # View recent interactions
/paperchat clear # Clear your chat history
Known Issues & Limitationsions
Beta Limitations
- Memory Storage Only: Conversation history is lost on server restart
- Single Server Support: No cross-server communication
- Basic Permission System: Only
paperchat.usepermission implemented
Fixed Issues from v1.0.0-beta
- Improved JSON response parsing reliability with sanitization library
- Resolved provider discovery issues with ServiceLoader
- Added a HackClub provider for free Hack Club members' AI access
- Enhanced command structure for more intuitive usage
- Resolved cross-player spam abuse
New Dependencies
com.mikesamuel:json-sanitizer:1.2.3 - For robust JSON response cleaning
Provider Comparison
| Provider | API Key Required | Models Available | Cost |
|---|---|---|---|
| HackClub | No | qwen/qwen3-32b, openai/gpt-oss-120b, others (selected server-side) | Free for teens enrolled in Hack Club |
| Yes | Gemini family models | Pay per use | |
| OpenAI | Yes | GPT family models | Pay per use |
License
Licensed under Apache 2.0 License - see LICENSE file for details.
Note: This is a beta release intended for testing and feedback. While stable for basic usage, some features may change or new ones be implemented in future releases.