```
# elizaOS Tweet Ideas

1. unveiling elizaos v2: a complete architectural overhaul from project-scoped to agent-scoped plugins - characters can now have their own unique plugin configurations

2. the venice plugin just got a major update (v1.0.13) with image support, text-to-speech, and no more openai dependency for embeddings

3. v1.0.11 is out with significant architectural improvements and memory management updates. the knowledge upload functionality is fixed for those who were experiencing issues

4. our message system has been completely rebuilt to be standalone from agents, making room for more flexible communication patterns in your multi-agent systems

5. we've been hard at work significantly improving plugin architecture to support per-agent configurations rather than global project settings

6. "v2 live docs" is coming soon - shaw has been heads down working on completing elizaos v2, which promises major improvements for ai agent developers

7. developers: we've added support for user-uploaded .txt files in the gui and improved how attachments are handled, including local image urls and descriptions

8. the team is making significant progress on knowledge management features - we understand this has been a pain point in recent versions

9. claude code integration has been added to our github workflow, enhancing our development process with ai-assisted code review and generation

10. we've added comprehensive testing infrastructure for core components, server functionality, and starter packages to ensure reliability as we move forward

11. just pushed a fix for agent cross-interference that was causing multiple agents to respond to messages intended for only one agent

12. attention plugin developers: we've made significant improvements to the plugin loading strategy to resolve core dependency conflicts

13. big news for crypto/web3 developers - we've added actions and abis for the alethea plugin to enhance web3 capabilities in elizaos

14. community update: q3 is expected to bring significant growth for ai agents, especially as elizaos v2 arrives with its architectural improvements

15. we've fixed real-time message deletion and added cascade delete for agents to improve your development workflow

# Concise Twitter Thread

1/ elizaos v2 development update: shaw is making significant progress on the next major version that fundamentally rethinks how agents and plugins interact. the core architecture is shifting from project-scoped to agent-scoped plugins.

2/ this means each character in your system can have its own unique plugin configuration rather than sharing a global set. this enables more specialized agents and clearer separation of capabilities between different characters in your multi-agent systems.

3/ alongside this, we've been improving our knowledge management system, fixing upload functionality, and enhancing our testing infrastructure. the venice plugin now supports images, text-to-speech and no longer requires openai for embeddings.

# Platform-specific Posts

## For ElizaOS's technical audience:
in the 1.0.11 release we've completely refactored our plugin architecture to support agent-scoped plugins rather than project-scoped plugins. this architectural shift enables developers to configure different plugins for different characters within the same project, creating more specialized agents with cleaner separation of concerns. we've also centralized directory detection logic and implemented a new type-safe api client package with comprehensive typescript support.

## For auto.fun's crypto-native audience:
gm devs. just shipped some massive upgrades to our web3 integrations. added actions and abis for the alethea plugin, did a full refactoring of polygon plugin to avoid naming conflicts, and fixed evm integration issues that were blocking agent-blockchain interactions. if you're building token-aware agents, this update makes it way easier to interact with on-chain data and execute transactions. full commit history on gh if you wanna dive deeper.
```