# elizaOS Discord - 2025-03-27

## Overall Discussion Highlights

### Project Updates & Announcements
- **Website Countdown**: The Eliza Studios website featured a countdown that completed today, leading to the launch of **auto.fun**, described as a launchpad.
- **Version Updates**: Users are transitioning between different versions (v0.25.9, v1.0.0, v2) with various migration challenges being discussed.
- **Spartan on Discord**: Team member rhota confirmed they're working on a v2 that will allow users to interact with Spartan on Discord soon, without waiting for X's review process.
- **Binance Integration**: Binance Alpha 2.0 now allows purchasing degenAI without requiring a web3 wallet.

### Technical Discussions
- **Local LLM Performance**: Users reported running llama3.1:8b-instruct-q4_K_M on consumer hardware (GeForce 3060 GPU with 12GB VRAM), achieving 15 tokens per second with a 32K token context window.
- **Plugin Integration**: Several users discussed configuration issues with Twitter, Discord, and custom plugins across different versions.
- **API Rate Limiting**: Discussion about Anthropic's limits (50k input tokens per minute) and strategies for managing them.
- **Character Configuration**: Significant discussion about the correct approach for creating and configuring character files in the newer v2 architecture.

### AI Image Generation Advancements
- **GPT-4o Capabilities**: Jin shared examples demonstrating GPT-4o's ability to handle multiple characters in a scene consistently (up to 5 characters).
- **Workflow Evolution**: Discussion highlighted how AI image generation has evolved from complex workflows requiring ControlNet IP-Adapter, ComfyUI, negative prompts, and Loras to a much simpler natural language interface.
- **Future Implications**: Patt compared this evolution to how smartphones transformed computing, suggesting tools like GPT-4o will similarly simplify and replace more complex AI image generation tools.

### Token & Financial Discussions
- **AI16Z Staking**: Users inquired about staking opportunities for AI16Z token, with responses indicating it's only available as a 2-sided LP with SOL on daos.fun.
- **Token Ticker**: There's an ongoing metadata upgrade to change the token ticker, with teams working together on this issue.

## Key Questions & Answers

### Technical Implementation
- **Q**: How do plugins work in the CLI and how to add a custom API?  
  **A**: For v0.25.9, add plugins to the character configuration file with the fully qualified plugin name in the "plugins" array.

- **Q**: What's the correct way for creating a new character in Eliza v1.0?  
  **A**: For v2, use the CLI as documented at https://eliza.how/docs/cli/agent

- **Q**: Where should I put the JSON character file in Eliza v2?  
  **A**: Use the CLI as documented at https://eliza.how/docs/cli/agent

- **Q**: What are the new imports for composeContext and generateObject from "@elizaos/core"?  
  **A**: Use composePrompt instead of context, and useModel(ModelType.OBJECT_SMALL)

### Project Status
- **Q**: Is it possible to stake ai16z anywhere?  
  **A**: Only as 2-sided LP with SOL on the daos.fun page for now.

- **Q**: What is auto.fun?  
  **A**: It's the launchpad.

- **Q**: When will we be able to speak with Spartan again on Discord or X?  
  **A**: It feels close but can't give an exact date due to rapid pace of change around v2. You will be able to interact with Spartan on Discord first, soon.

- **Q**: Should we get the autodot.fun account verified?  
  **A**: It's under review.

### AI Capabilities
- **Q**: Would you be able to generate the rest while keeping the style consistent with the initial eliza & Marc image?  
  **A**: It starts to struggle past 5+ chars in a scene, but that alone is insane progress.

- **Q**: How did previous character generation workflows compare to current ones?  
  **A**: Previous methods required ControlNet IP-Adapter, ComfyUI, negative prompts, and Loras with sophisticated prompts, but now it's mind-bogglingly simple.

## Community Help & Collaboration

### Plugin Configuration
- **Helper**: kaisermerkle | **Helpee**: Prince  
  **Context**: How to configure Twitter client in character.json  
  **Resolution**: Provided JSON example showing how to add Twitter plugin to the "plugins" array instead of using the deprecated "clients" key

### Local AI Configuration
- **Helper**: winded4752 | **Helpee**: Abderahman  
  **Context**: Explaining what localAiPlugin is  
  **Resolution**: Explained it's for Ollama/LMStudio or local to Eliza's DB directory

### Documentation & Migration
- **Helper**: jin | **Helpee**: Multiple users  
  **Context**: Documentation for character creation in v2  
  **Resolution**: Shared documentation link and updated GitHub repo description to make docs more discoverable

- **Helper**: new.moon | **Helpee**: kaisermerkle  
  **Context**: Updated import syntax for core functions  
  **Resolution**: Provided correct imports (composePrompt instead of composeContext, useModel(ModelType.OBJECT_SMALL))

### Installation Support
- **Helper**: winded4752 | **Helpee**: Multiple users  
  **Context**: Installation issues with v2  
  **Resolution**: Suggested cloning the v2-develop repo, then running 'bun install' and 'bun run build'

### Token Information
- **Helper**: Patt | **Helpee**: rogerdefi  
  **Context**: Asking about staking opportunities for AI16Z  
  **Resolution**: Informed that it's only available as 2-sided LP with SOL on daos.fun

- **Helper**: Patt | **Helpee**: Poloethr  
  **Context**: Inquiring about progress on ticker change  
  **Resolution**: Explained it's a metadata upgrade requiring development with ongoing dialogue between teams

## Action Items

### Technical
- Fix Twitter client in v2 to handle duplicate status errors properly (Mentioned by Abderahman)
- Resolve import errors with @elizaos/plugin-sql and @elizaos/plugin-local-ai (Mentioned by kaisermerkle)
- Fix Telegram client image processing with OpenRouter/DeepSeek (Mentioned by kaisermerkle)
- Fix Discord package compatibility with v2 beta (Mentioned by gourshanne)
- Fix issue with Twitter plugin on v1 that tweets unrelated content (Mentioned by RaglioKen)
- Change token ticker (metadata upgrade) (Mentioned by Poloethr)
- Renounce the smart contract (Mentioned by Poloethr)
- Fix agents showing as disconnected in version 0.25.9 (Mentioned by 21givenchy)
- Implement v2 of Spartan for Discord (Mentioned by rhota)
- Explore open-source alternatives to OpenAI's image generation technology (Mentioned by jin)

### Documentation
- Improve documentation visibility and SEO for eliza.how (Mentioned by jin)
- Create migration guide from v0.25.9 to v1.0.0 (Mentioned by kaisermerkle)
- Update character file documentation for v2 (Mentioned by Abderahman)
- Write a guide for the new image generation technique (Mentioned by jin)
- Review and provide feedback on the "taming_info" blog post (Mentioned by jin)

### Feature
- Add support for more cost-effective Twitter model providers (Mentioned by Creed)
- Add tracing capabilities for LLM interactions similar to LangSmith (Mentioned by ad0ll)
- Consider rebranding with a new character/account for Spartan (Mentioned by 쭈리남푠, kkeongsmemo)
- Improve character consistency in multi-character scenes beyond 5+ characters (Mentioned by jin)