{
  "ai-news": {
    "elizaos": {
      "discord_md_last_3_days": [
        {
          "filename": "2025-04-17.md",
          "content": "# elizaOS Discord - 2025-04-17\n\n## Overall Discussion Highlights\n\n### Auto.fun Platform Launch\n- **Launch Timeline**: The auto.fun platform is expected to launch \"this week,\" possibly today, with final testing underway\n- **No Pre-announcement Strategy**: Team prefers not to pre-announce exact launch dates to concentrate messaging when users can immediately try the platform\n- **Platform Features**: Will allow users to create coins through AI prompts, with customizable token suffixes (defaulting to \"fun\")\n- **Plugin Support**: Removed from initial release due to complexity for non-technical users; will be supported in a future agent builder product\n- **Differentiation**: Aims to provide better security features than competitors like pump.fun and return value to the ecosystem\n\n### Token Confusion & Controversy\n- **Unofficial Token Issue**: Significant confusion arose when Shaw purchased a token called \"auto.fun\" on pump.fun shortly before the official platform launch\n- **Security Concerns**: The unofficial token's dexscreener link contained malicious code (drainer)\n- **Official Tokens**: Team repeatedly clarified that only $ai16z and $degenai are official tokens\n- **Community Reaction**: Partners expressed concerns about negative perception and potential harm to users from the unofficial token situation\n\n### ElizaOS Technical Issues\n- **V2 Beta Status**: ElizaOS V2 is in beta with many users experiencing installation and configuration difficulties\n- **Common Problems**: Issues with dependencies, environment variables, plugin compatibility, and Docker container setup\n- **Windows Compatibility**: Docker builds failing due to CRLF line ending issues on Windows\n- **Documentation Gaps**: Tutorials may be outdated for V2, causing setup confusion\n\n### DAO Communication Improvements\n- **Communication Concerns**: Partners expressed frustration about lack of regular updates from the team\n- **Monthly Town Hall**: First Labs/DAO sync scheduled for May 1st after 4pm PST / 7pm EST\n- **Format**: Will collect questions from community beforehand to make efficient use of time\n- **Transparency**: COO acknowledged communication issues and committed to improvement\n\n## Key Questions & Answers\n\n### Auto.fun Platform\nQ: When is auto.fun launching?  \nA: This week, with no specific pre-announcement of time/date.\n\nQ: Is the auto.fun token (on pump.fun) official?  \nA: No, only $ai16z and $degenai are official tokens.\n\nQ: Will auto.fun have plugin support out of the box?  \nA: No, plugins created too many edge cases for non-technical users and will be supported in a more serious agent builder product later.\n\nQ: Can tokens issued through auto.fun have custom CA suffixes?  \nA: Yes, users can pick the suffix they want, with three-letter suffixes generating quickly. It likely defaults to \"fun\".\n\n### ElizaOS Technical\nQ: Where should the .env file be created?  \nA: In the root of the project.\n\nQ: How to set different models for different agents?  \nA: Create different folders with different env files with different models.\n\nQ: How do I change my postgres database once setup?  \nA: It's in ~/.eliza/.env - use cat to see it, remove it with editor, then start again.\n\nQ: Is v2 stable to use for production?  \nA: No, it doesn't work with the plugin registry.\n\n### DAO & Communication\nQ: Why hasn't the team been providing regular updates to the community?  \nA: The COO has been focused on operations, accounting, legal, and financials, but acknowledges the need for better communication, especially from the product team.\n\nQ: Why hasn't the team been promoting the $ai16z token?  \nA: The team plans to release detailed tokenomics for ai16z with autofun at launch.\n\nQ: Will there be a recording or notes from the town hall?  \nA: Yes, they will provide recordings or notes.\n\n## Community Help & Collaboration\n\n### ElizaOS Technical Support\n- **Docker Configuration**: .0xbbjoker helped de_fi_ne with Docker build failures by providing working Docker commands and configuration for PostgreSQL connection\n- **Windows Line Endings**: mevinator identified that Windows CRLF line endings needed to be converted to Unix LF for Docker builds\n- **Database Issues**: .0xbbjoker advised cocaine7499 to remove ~/.eliza directory and start again to resolve database connection issues\n- **OpenAI Module Failures**: samdeveloper suggested cleaning installation with \"bun run clean, bun install, bun build\" to fix module loading problems\n\n### Security Warnings\n- **Drainer Alert**: Smore and yikesawjeez repeatedly warned users about the drainer link in the dexscreener for the unofficial auto.fun token\n- **Comment Warnings**: yikesawjeez posted warnings in the comments section of relevant posts to protect users\n\n### Community Organization\n- **Town Hall Planning**: vincentpaul helped accelxr organize the first Labs/DAO town hall, proposing the May 1st date and suggesting a format for collecting questions beforehand\n- **Chinese Media Outreach**: Zolo shared links to news coverage about AI16z in Chinese media to help promote the project\n\n## Action Items\n\n### Technical\n- Launch the auto.fun platform this week (shaw)\n- Fix plugin-sql version compatibility issues with v2 (mevinator)\n- Resolve embedding provider requirements when using Anthropic without OpenAI (mevinator)\n- Fix character import functionality for JSON files (mevinator)\n- Address Telegram plugin chat behavior inconsistencies (Okul)\n- Fix Twitter agent timing configuration (0xCryptoCooker)\n- Implement Twitter image/video posting capability (mevinator)\n- Fix Windows compatibility for Docker builds (CRLF issue) (mevinator)\n- Fix V2 setup process to make it easier for users (Fog)\n- Ensure auto.fun server can handle traffic at launch (HoneyBadger)\n- Fix scoring and XP algorithms for user profiles (jin)\n- Add security features to scan dexscreener links for malicious code (yikesawjeez)\n\n### Documentation\n- Update Eliza website to include information about degenai (ɱɑყɑɱɑεʂƚɾ)\n- Update tutorials for V2 as current ones are outdated (Kenk)\n- Create clear documentation for auto.fun (Multiple users)\n- Update CLI documentation to mention bun requirement on Mac (mevinator)\n- Document environment variable requirements and locations (mevinator)\n- Add migration guide for v1 plugins to v2 (nibbles0737)\n- Document how to set different models for different agents (subhankar141202)\n- Prepare detailed tokenomics post for ai16z with autofun for launch (accelxr)\n- Provide weekly updates to the community about product development (찌 G 跻 じ PrudentSpartan)\n- Record or provide notes from the town hall for those who cannot attend (accelxr)\n- Create how-to reels/shorts and TikTok content for auto.fun (𝔭𝔩𝔞𝔱𝔞 𝔑𝔬 𝔉𝔞𝔭 𝔞𝔯𝔠)\n- Add disclaimer about official tokens on auto.fun website (Curtisdonthurtus)\n\n### Feature\n- Support for setting different models per agent (subhankar141202)\n- Improve Twitter agent with image/video posting capabilities (mevinator)\n- Add ability to disable admin panel for security (wookosh)\n- Implement auto-moderation to remove spam/solicitation messages (Osint)\n- Restrict posting links to mods/team members only (Osint)\n- Clarify on auto.fun X account that there are only two tokens ($ai16z and $degen) to prevent confusion (yikesawjeez)\n- Organize monthly sync between Labs and DAO starting May 1st (vincentpaul)\n- Build an AI agent that acts as a DAO delegate (cattin)\n- Develop a more serious agent builder product with plugin support (eskender.eth)\n- Implement security scanning for dexscreener links on auto.fun (yikesawjeez)"
        },
        {
          "filename": "2025-04-16.md",
          "content": "# elizaOS Discord - 2025-04-16\n\n## Overall Discussion Highlights\n\n### Auto.fun Launch Preparation\n- **Launch Timeline**: Auto.fun platform confirmed to launch \"this week\" with several partners lined up for post-launch activities\n- **Platform Features**: \n  - AI-generated meme tokens\n  - Client-side vanity contract address generation (users can customize suffixes)\n  - Integration with Raydium for liquidity pools\n  - Creators will receive NFTs from Raydium to claim creator fees\n  - Day 1 community token creation capability\n- **Token Confusion**: Controversy around a fake \"auto.fun\" token on pump.fun that Shaw purchased shortly before the official platform launch\n- **Revenue Model**: Platform revenue will contribute to buying back AI16z tokens, though specific economics haven't been finalized\n\n### ElizaOS Development\n- **Version Confusion**: Users experiencing issues with different versions:\n  - V2 (referred to as 1.x series) is in beta\n  - V1 (referred to as 0.x series) is the stable version\n  - Tutorials appear to be outdated for V2 beta\n- **Plugin Issues**: Multiple users reporting problems with plugin installation and configuration:\n  - Plugin loading failures with @elizaos/plugin-openai and @elizaos/plugin-sql\n  - Environment variable configuration issues\n  - Package management differences between npm, pnpm, and bun\n- **Documentation Problems**: Breaking changes pushed to documentation requiring fixes\n- **Leaderboard System**: Temporarily down but restored with backend improvements at elizaos.github.io/profile/\n\n### Communication Improvements\n- **DAO/Labs Coordination**: Discussion about improving communication between Eliza Labs and the DAO community\n- **New Initiatives**:\n  - Monthly Labs/DAO town hall scheduled for May 1st\n  - Additional social media support being added\n  - Brand communications kit in development\n  - Rotational developer support implementation\n  - Structured internal update processes being established\n- **Knowledge Repository**: Automation implemented to keep GitHub knowledge repository updated daily\n\n## Key Questions & Answers\n\n### Auto.fun Platform\n- **Q**: Is it a platform launch or TGE? **A**: Platform launch (Odilitime)\n- **Q**: Token launch same day? **A**: There is no new token. We're just $ai16z and $degenai (Odilitime)\n- **Q**: What kind of volume are you expecting comparing to pump.fun? **A**: Better than flaunch but still orders of magnitude off from pump.fun (shaw)\n- **Q**: So utility only coins? Or can there be memes? **A**: Definitely memes. You can go to /create and press launch with AI-generated meme content ready to go (shaw)\n- **Q**: Can the tokens issued through auto.fun have CA suffixes like elizaOS or fun? **A**: Users can pick the suffix they want and generate it. Three letter suffixes are really quick (accelxr)\n- **Q**: Will people be able to create tokens out the gate? **A**: Yes, day 1 community creation (eskender.eth)\n- **Q**: Part of the revenue from auto.fun goes toward buying back ai16z, is it right? **A**: The plan is to make the number go up, but specific economics haven't been worked out (shaw)\n\n### ElizaOS Technical\n- **Q**: What is v2? Isn't v2 only in develop-v2? If this is called 1.0.0-beta that is also v2? **A**: V2 is 1.x, v1 is 0.x (cocaine7499)\n- **Q**: How can I start ElizaOS properly when guides aren't up to date? **A**: Try using v0.25 with the openai plugin. For v1.0x, we'll let the community know when plugins have been migrated. (Kenk)\n- **Q**: How can I change my postgres database once setup? **A**: It's in your ~/.eliza/.env: cat ~/.eliza/.env to see, open with vim or code editor remove it, start again (.0xbbjoker)\n- **Q**: What is the best way to run my bot continuously? **A**: You can use pm2, but you might need to reconnect your SSH session after installing it (cocaine7499)\n- **Q**: Does anyone have a potential example of using Google Gemini? **A**: We don't have a way for google/gemini models yet. You can use plugin-openai, plugin-anthropic, plugin-groq, or plugin-venice instead. (cocaine7499)\n\n### Community and Organization\n- **Q**: Will we have a recording of the town hall or notes? **A**: Yes, of course (accelxr)\n- **Q**: What should be the focus for partners? **A**: Support the auto.fun launch rather than broader dao-org push (vincentpaul)\n\n## Community Help & Collaboration\n\n### ElizaOS Technical Support\n- **cocaine7499** helped users understand available plugins for different models and how to set them up with environment variables\n- **Kenk** assisted **.trbr** by suggesting PDF Extraction Plugin and Firecrawl Plugin as alternatives to OpenAI plugin for PDF processing\n- **samdeveloper** provided solutions for plugin installation issues, suggesting \"bun run clean, bun install, bun build\" sequence and step-by-step instructions to clone the repository directly using the v2-develop branch\n- **.0xbbjoker** helped **cocaine7499** change postgres database configuration by explaining how to find and edit the configuration in ~/.eliza/.env\n- **Kenk** helped **Fog** with ElizaOS setup issues by explaining they're in V2 beta and tutorials might be outdated, advising to use v2 develop branch not main\n\n### Knowledge Sharing\n- **jin** implemented automation to keep GitHub knowledge repository updated daily\n- **yikesawjeez** demonstrated how to use the community dashboard to filter for topics of interest\n- **shaw** shared screenshots and details about auto.fun features including AI-generated content, vanity addresses, and Raydium integration\n\n### Community Coordination\n- **vincentpaul** proposed and organized a monthly town hall starting May 1st\n- **HoneyBadger** offered to introduce Nikita Bier to shaw for UX advice on auto.fun\n- **accelxr** outlined comprehensive communication improvements and requested partner help with auto.fun launch\n\n## Action Items\n\n### Technical\n- Fix plugin loading issues in v1.0.0-beta, particularly for plugin-openai and plugin-sql (mentioned by cocaine7499)\n- Implement support for Google/Gemini models (mentioned by cocaine7499)\n- Fix metadata ownership requirement causing \"null is not an object (evaluating 'userWorlds.find')\" error (mentioned by cocaine7499)\n- Fix Telegram plugin action triggering inconsistencies (mentioned by Okul)\n- Launch auto.fun platform this week as scheduled (mentioned by shaw)\n- Implement API for trading terminals post-launch (mentioned by eskender.eth)\n- Continue frontend improvements for the leaderboard system (mentioned by jin)\n- Fix scoring and XP algorithms for the leaderboard (mentioned by jin)\n- Integrate with Fleek to host Eliza agents (mentioned by shaw)\n- Fix documentation after breaking change (mentioned by jin)\n- Address model content reuse issue (mentioned by Waqas Wahid)\n- Address wallet drainer link issue (mentioned by yikesawjeez)\n- Monitor and moderate the auto.fun channel during launch (mentioned by accelxr)\n\n### Documentation\n- Update guides to clarify differences between v0.x and v1.x versions (mentioned by .trbr)\n- Create clearer documentation about environment variable placement and configuration (mentioned by cocaine7499)\n- Provide examples of using different model providers with ElizaOS (mentioned by cocaine7499)\n- Create guide for feeding knowledge to characters through PDFs (mentioned by .trbr)\n- Update documentation for running bots continuously with PM2 (mentioned by cocaine7499)\n- Update tutorials for V2 (mentioned by Kenk)\n- Prepare detailed tokenomics post on ai16z with autofun for launch (mentioned by accelxr)\n- Implement weekly updates from Labs to DAO (mentioned by PrudentSpartan)\n- Collect questions from community ahead of town hall (mentioned by vincentpaul)\n- Publish detailed tokenomics and economic model for auto.fun (mentioned by Zolo)\n- Create assets of the autofun mascot or other on-brand assets for launch day (mentioned by accelxr)\n- Share launch threads for X with accelxr or team for amplification (mentioned by accelxr)\n\n### Feature\n- Implement auto-deletion of spam messages (mentioned by Osint)\n- Restrict link posting permissions (mentioned by Osint)\n- Enable agents to communicate with each other outside of Discord (mentioned by cocaine7499)\n- Implement vampire attack features and AI takeover for auto.fun (mentioned by shaw)\n- Add more partners beyond Raydium for LP pools (mentioned by shaw)\n- Develop Eliza v2 shortly after auto.fun launch (mentioned by shaw)\n- Purchase verification checkmark for Spartan page (mentioned by A)"
        },
        {
          "filename": "2025-04-15.md",
          "content": "# elizaOS Discord - 2025-04-15\n\n## Overall Discussion Highlights\n\n### Auto.fun Launch Preparation\n- Auto.fun platform is preparing to launch \"this week\" after some delays\n- UI screenshots were shared showing features like client-side vanity contract address generation, AI-generated token content (names, descriptions, images), and Raydium integration\n- The platform aims to compete with pump.fun but with unique AI features\n- Revenue from Auto.fun will contribute to buying back ai16z tokens\n- Day 1 will allow community token creation\n- Expected volume to be \"better than flaunch but still orders of magnitude off from\" pump.fun\n- Users will be able to connect agents to tokens on the platform\n\n### ElizaOS Version Compatibility Issues\n- Users are experiencing difficulties with the transition between ElizaOS versions (V1/0.x and V2/1.x)\n- Plugins from V1 are not compatible with V2 yet as they haven't been migrated\n- Several users reported issues with the OpenAI plugin in V2/beta versions\n- Configuration challenges include environment variables not being recognized properly and embedding model limitations\n\n### Model Provider Integration\n- Users are trying to configure different AI models including OpenRouter, Ollama, and Google Gemini\n- Discussion about plugin-ollama, plugin-openai, plugin-anthropic, plugin-groq, and plugin-venice with appropriate .env configuration\n- Some users working on developing plugin-openrouter integration for accessing various model categories\n\n### DegenAI Development\n- DegenAI is being actively developed, with Odilitime identified as one of the developers\n- Plans for promotion in conjunction with Auto.fun launch\n- Project combines AI, tech, and \"memetic value\"\n\n### Communication Improvements\n- Criticism of poor communication around launch delays and lack of updates\n- Comprehensive improvements being implemented including:\n  - Additional social media support\n  - Brand communications kit\n  - Improved documentation\n  - Rotational developer support\n  - Structured update processes\n  - DAO bounty board\n\n## Key Questions & Answers\n\n**Q: When is Auto.fun launching?**  \nA: \"This week\" (answered by shaw)\n\n**Q: Does part of the revenue from auto.fun go toward buying back ai16z?**  \nA: \"The plan is definitely to make the number go up, as that is the quickest way for us to all realize our bags\" (answered by shaw)\n\n**Q: Will people be able to create tokens out the gate?**  \nA: \"Yeah day 1 community creation\" (answered by eskender.eth)\n\n**Q: Are plugins from V1 compatible with V2/1.x?**  \nA: No, you can't use V1 plugins with V2 yet (answered by odilitime)\n\n**Q: Why is the OpenAI plugin not loading in V2/beta?**  \nA: The plugins are yet to be migrated to V2, which will happen when V2 is released (answered by Kenk)\n\n**Q: Is there documentation for the Ollama plugin?**  \nA: Yes, in the README at github.com/elizaOS/eliza/tree/v1.0.0-beta.32/packages/plugin-ollama (answered by cardinal.error)\n\n**Q: Is there any potential support for utility agents that want to launch on autofun?**  \nA: \"Would love to connect, can't speak explicitly to auto.fun support but would love to learn more about what you've built\" (answered by Kenk)\n\n**Q: Is it a platform launch or TGE?**  \nA: Platform launch (answered by Odilitime)\n\n**Q: token launch same day?**  \nA: \"There is no new token. We're just $ai16z and $degenai\" (answered by Odilitime)\n\n## Community Help & Collaboration\n\n### Technical Support\n- **.0xbbjoker helped cardinal.error** with a Postgres plugin error related to \"levenshtein argument exceeds maximum length of 255 characters\" by sharing a PR fix that limits the length of text for levenshtein comparison\n- **chris.troutner offered assistance** to users having issues with getting Ollama working with beta.32\n- **Kenk helped .trbr** who needed alternatives to OpenAI plugin for PDF processing by suggesting PDF Extraction Plugin, Firecrawl Plugin, and Bio Protocol's approach using V2 beta's API capabilities\n\n### Documentation Sharing\n- **odilitime helped pupathebig** looking for character file documentation by sharing the link: eliza.how/docs/0.25.9/core/characterfile\n- **Borko shared a Notion document** with platform details and mechanics about Auto.fun\n- **jin implemented daily updates** for the knowledge repository with automation\n\n### Prompt Engineering\n- **yikesawjeez helped Odilitime** with AI prompt engineering by sharing a tip: \"proceed as your docs, taskmaster, and other rules indicate you should\"\n\n### Community Contributions\n- **Void volunteered** to help with quantitative finance aspects of the Spartan project\n- **accelxr shared comprehensive updates** on communication improvements being implemented and requested specific partner assistance\n\n## Action Items\n\n### Technical\n- Migrate plugins from V1 to V2 (Mentioned by Kenk)\n- Fix OpenAI plugin loading issue in V2/beta (Mentioned by cardinal.error)\n- Develop plugin-openrouter integration (Mentioned by .0xbbjoker)\n- Fix Postgres plugin levenshtein limitation (Mentioned by .0xbbjoker)\n- Improve Telegram client to avoid duplicate messages (Mentioned by pupathebig)\n- Implement API for trading terminals post-launch (Mentioned by eskender.eth)\n- Filter out AI refusal responses from Spartan bot (Mentioned by shaw)\n- Consider moving to Deno for improved security (Mentioned by yikesawjeez)\n- Integrate Immunefi security solutions for the platform (Mentioned by yikesawjeez)\n- Fix documentation after breaking change (Mentioned by jin)\n- Resolve model content reuse issue (Mentioned by Waqas Wahid)\n- Monitor and moderate the auto.fun channel during launch (Mentioned by accelxr)\n- Implement integration with GMGN team for agent tokens on their platform (Mentioned by Zolo)\n\n### Documentation\n- Create updated guide for starting ElizaOS (Mentioned by .trbr)\n- Clarify character file configuration (Mentioned by pupathebig)\n- Improve clarity about version naming (V1/0.x vs V2/1.x) (Mentioned by cardinal.error)\n- Create detailed tokenomics documentation for Auto.fun (Mentioned by HoneyBadger)\n- Establish a single source of truth for community information (Mentioned by yikesawjeez)\n- Create assets of the auto.fun mascot for launch day (Mentioned by accelxr)\n- Prepare launch threads for X to amplify auto.fun launch (Mentioned by accelxr)\n- Clarify the situation regarding auto.fun delays to the community (Mentioned by Zolo)\n- Create bounty board to list needs and facilitate connections (Mentioned by Kenk)\n\n### Feature\n- Enable inter-agent communication outside Discord (Mentioned by cardinal.error)\n- Add authentication middleware for REST API endpoints (Mentioned by narasim)\n- Implement retroactive public goods funding (retropgf) (Mentioned by yikesawjeez)\n- Add Magnus upsell at the end of Auto.fun process (Mentioned by yikesawjeez)\n- Develop agents that people actually use, not just talk about using (Mentioned by shaw)\n- Implement more vampire attack features and AI takeover capabilities (Mentioned by shaw)\n- Support for utility agents on Auto.fun (Mentioned by Amir)\n- Establish partners + labs syncs for better coordination (Mentioned by accelxr)\n- Purchase verification checkmark for Spartan social media page (Mentioned by A)"
        }
      ],
      "dev_md_last_3_days": [
        {
          "filename": "2025-04-17.md",
          "content": "# elizaOS Development Discord - 2025-04-17\n\n## Overall Discussion Highlights\n\n### ElizaOS v2 Development\n- The community is actively working with the v2 beta, developing plugins, and adapting to architectural changes\n- Most plugins are still in transition from v1 to v2\n- Client architecture has changed significantly in v2, with the previous client system being replaced by a plugin-based architecture\n- Knowledge directory structure has changed in v2, with files now typically stored under projects/[project_name]/knowledge/\n- API endpoints have been restructured in v2 under /api/v2/ with separate routes for agents, projects, and knowledge management\n\n### Plugin Development\n- Developers are working on integrating various plugins into the ElizaOS v2 ecosystem, including:\n  - Discord plugin\n  - Twitter plugin\n  - Suno (music AI) plugin\n  - MCP plugin by Fleek\n  - Livepeer as an inference provider for Vtuber plugin\n- Discussions about how to add and use local plugins in the monorepo environment without publishing them\n- Questions about adding new inference providers under the packages folder\n\n### Technical Challenges\n- PDF upload functionality is not supported in v2, requiring manual file placement in knowledge directories\n- For production environments with multiple characters, separate projects are recommended for better isolation and control\n- A workaround for PDF handling was suggested using GPT-4's PDF reading capability combined with a custom embedding pipeline\n\n### Pull Requests & Code Reviews\n- PR #100 for the agent-twitter-client repo implementing tweet deletion feature\n- PR #4320 for plugin-twitter to maintain consistency across codebases\n- PRs for XMTP client upgrades (#2) and registry updates (#127)\n\n## Key Questions & Answers\n\n**Q: How do I import a local plugin in monorepo and use it in defaultCharacter?**  \nA: You should be able to add the plugin in monorepo, then build it and reference it in the character file. (0xbbjoker)\n\n**Q: Can I find the Discord V2 plugin for my project?**  \nA: The v2-develop version should work with the beta version. (Nisita)\n\n**Q: Any good tutorial on how to create a client?**  \nA: Clients are not in v2 anymore; check existing clients in v1 like Auto-client and direct-client. (standard)\n\n**Q: Can the agents or the org use MCP?**  \nA: Yes, there's a plugin-mcp by fleek. (Odilitime)\n\n**Q: Where is the knowledge directory in the new rework?**  \nA: In v2 the knowledge directory structure changed - check the config/projects folder for the default path, or specify a custom knowledge_path in your agent config. (Ruby)\n\n**Q: For production, is it better to use multiple projects for each character or just one?**  \nA: Multiple projects gives better isolation and control over each character's knowledge and behavior. Single project is simpler but requires careful namespacing. (Ruby)\n\n**Q: Does the project version have an API to upload knowledge?**  \nA: Not yet - knowledge upload via API is still in development. Files need to be added manually to project directories for now. (Ruby)\n\n**Q: Can I change the ID of an already created avatar?**  \nA: Yes, you can update an avatar's ID in the config file, but it will break existing conversations or knowledge tied to that ID. (Ruby)\n\n**Q: Could you return the response from the deleteTweet method?**  \nA: Thank you for the feedback, I'm adding the return to the wrapping method, that was an oversight. (4nzn)\n\n## Community Help & Collaboration\n\n### Plugin Development Support\n- **0xbbjoker** helped **standard** with adding and using local plugins in monorepo by providing step-by-step instructions for building, linking, and using local plugins\n- **standard** assisted **LemonS** with creating a client when no plugin exists by suggesting combining Auto-client and direct-client from v1 as a starting point\n- **Nisita** helped **モーテンセン** find the Discord V2 plugin and confirmed the v2-develop version works with beta\n\n### PDF Handling Solutions\n- **Ruby** provided **mindxploit** with a comprehensive solution for PDF RAG with v0.25.9, including code examples using langchain with PDFPlumberLoader, RecursiveCharacterTextSplitter, OpenAIEmbeddings, and FAISS to create a custom PDF processing pipeline\n\n### Code Review & Improvement\n- **tcm390** helped **4nzn** improve the deleteTweet method implementation by suggesting returning the Response object and opening a PR to the v2-develop branch for consistency\n- **Ruby** reviewed and approved **4nzn**'s PR after confirming code quality, test coverage, and API integration\n\n## Action Items\n\n### Technical\n- Implement custom PDF handling via GPT-4 and embedding pipeline for v0.25.9 (Mentioned by Ruby)\n- Export important data before changing agent IDs (Mentioned by Ruby)\n- Add Response return type to deleteTweet method in agent-twitter-client (Mentioned by tcm390)\n- Create PR for tweet deletion feature in v2-develop branch of plugin-twitter (Mentioned by tcm390)\n- Review XMTP client upgrades and registry updates (Mentioned by Ruby)\n- Add Livepeer as inference provider for Vtuber plugin (Mentioned by DeFine)\n- Port client functionality into the new plugin-based architecture (Mentioned by standard)\n\n### Documentation\n- Create better tutorials for plugin development in v2 (Mentioned by LemonS)\n- Create clear documentation for agent ID configuration (Mentioned by mindxploit)\n\n### Feature\n- Add PDF upload support to v2 and projects mode (Mentioned by mindxploit)\n- Develop API endpoints for knowledge management (Mentioned by Ruby)\n- Add joystram to Eliza (Mentioned by Yemmii)\n- Integration of Suno music AI plugin (Mentioned by DorianD)\n- Implement Discord plugin for v2 (Mentioned by モーテンセン)"
        },
        {
          "filename": "2025-04-16.md",
          "content": "# elizaOS Development Discord - 2025-04-16\n\n## Overall Discussion Highlights\n\n### ElizaOS v1 to v2 Migration\n- Users are actively migrating from v1 to v2, with several reporting challenges during the transition\n- The monorepo structure in v2 requires different approaches for plugin integration and development\n- Several v1 plugins are yet to be migrated to v2, including the Discord plugin\n- Clients are handled differently in v2 compared to v1, with some users seeking guidance on implementation\n\n### Database and Infrastructure\n- Discussion about database options with Postgres, Supabase, and Neon being considered\n- Supabase emerged as a preferred solution for many developers\n- Hidden Forces reported switching to Postgres and v2 for their project\n\n### Agent Inference and API Configuration\n- Detailed discussions about plugin order behavior affecting which models are registered and used first\n- Issues with agent crashes when switching between local-AI and cloud providers like OpenAI and Anthropic\n- System appears to default to Anthropic even when OpenAI keys are provided\n- API key configuration in .env files was a common troubleshooting topic\n\n### PDF Processing Challenges\n- PDF handling is problematic in both older versions (0.25.x) and newer v2/projects mode\n- Users experiencing errors like \"Invalid array length\" when using PDFs as knowledge sources\n- Ruby suggested workarounds including manual PDF addition and custom processing pipelines\n- Custom solution using GPT-4's PDF reading capabilities combined with embedding pipeline was proposed\n\n## Key Questions & Answers\n\n**Q: Do you use pglite or postgres? Which eliza version v1 or v2?**  \nA: Hidden Forces indicated switching to Postgres and v2 for their project\n\n**Q: How do I fix my agent crashing when typing to it in localhost?**  \nA: 0xbbjoker explained it was due to local-ai plugin issues and recommended removing ~/.eliza directory and ensuring proper API keys in .env\n\n**Q: How do I specify the OpenAI key or is it better to go with Anthropic?**  \nA: DeFine suggested using Anthropic for now and adding the key to .env file\n\n**Q: How do I import a local plugin in monorepo for use in defaultCharacter?**  \nA: 0xbbjoker explained the process of adding plugins in monorepo, building, and linking\n\n**Q: Where can I find the Discord V2 plugin?**  \nA: Nisita explained V2 plugins are yet to be migrated but the v2-develop version works with beta\n\n**Q: Any good tutorial on how to create a client?**  \nA: standard explained clients aren't in v2 anymore and suggested checking Auto-client and direct-client in v1\n\n**Q: Isn't PDF processing natively supported by Eliza knowledge?**  \nA: Yes, Eliza should handle PDF parsing natively through the knowledge pipeline, but there are issues with the implementation.\n\n**Q: Where is the knowledge directory in the new version?**  \nA: In v2, check the config/projects folder for the default path, or specify a custom knowledge_path in agent config.\n\n**Q: For production, is it better to use multiple projects for each character or just one?**  \nA: Depends on use case - multiple projects provide better isolation and control, single project is simpler but requires careful namespacing.\n\n**Q: Does the project version have an API to upload knowledge?**  \nA: Not yet - knowledge upload via API is still in development. Files must be added manually to project directories for now.\n\n## Community Help & Collaboration\n\n### Agent Troubleshooting\n- **0xbbjoker helped Cocaine** with agent crashes in local environment by explaining it was due to local-ai plugin issues, suggesting removal of ~/.eliza directory and proper API key configuration\n- **0xbbjoker provided Cocaine** with detailed explanation of plugin order behavior and how it affects model selection\n- **DeFine assisted Cocaine** with agent crashes by explaining the need to add Anthropic API key as the system defaults to Anthropic\n\n### Plugin and Development Support\n- **0xbbjoker guided standard** through importing local plugins in monorepo with step-by-step instructions for adding, building, and linking plugins\n- **standard helped LemonS** understand client creation by explaining clients aren't in v2 and suggesting combining Auto-client and direct-client from v1\n\n### PDF Processing Solutions\n- **Ruby helped mindxploit** diagnose PDF parsing errors, identifying corrupted file structure as the likely cause\n- **Ruby provided mindxploit** with code example using LangChain components to create a custom PDF processing pipeline compatible with 0.25.9\n- **Ruby explained to mindxploit** the basics of creating separate knowledge directories for each agent in v2\n\n## Action Items\n\n### Technical\n- Remove ~/.eliza directory when switching from local-ai to cloud providers (Mentioned by 0xbbjoker)\n- Fix bug where system errors when Anthropic API key is missing even when OpenAI key is provided (Mentioned by DeFine)\n- Add Livepeer as inference provider for Vtuber plugin (Mentioned by DeFine)\n- Implement custom PDF processing using GPT-4's PDF reader and custom embedding pipeline (Mentioned by Ruby)\n- Check PDF file structure integrity to resolve \"Invalid array length\" error (Mentioned by Ruby)\n- Break large PDF documents into chunks for better processing (Mentioned by Ruby)\n\n### Documentation\n- Create clearer documentation on plugin order behavior and API key configuration (Mentioned by 0xbbjoker)\n- Provide tutorial for creating custom clients in v1 (Mentioned by LemonS)\n- Improve documentation on setting up multiple agents with their own knowledge bases in v2 (Mentioned by mindxploit)\n\n### Feature\n- Add joystram integration to Eliza (Mentioned by Yemmii)\n- Migrate v1 plugins to v2 architecture (Mentioned by Nisita)\n- Implement endless conversations between two AI agents (Mentioned by desu)\n- Add PDF upload support in v2 and projects mode (Mentioned by mindxploit)\n- Develop API endpoints for knowledge management including file uploads (Mentioned by Ruby)"
        },
        {
          "filename": "2025-04-15.md",
          "content": "# elizaOS Development Discord - 2025-04-15\n\n## Overall Discussion Highlights\n\n### ElizaOS Development & Versions\n- Discussions about ElizaOS v1 and v2 implementation issues, particularly around plugin compatibility\n- Users reported challenges migrating from v1 to v2, especially when setting up multiple agents with separate knowledge bases\n- GPT-4.1 support is not yet available in elizaos/core v0.25.9\n- The \"the-org\" package was confirmed to be a demo app that starts up example agents for organizational use cases\n\n### Plugin & Integration Issues\n- Several users encountered plugin loading problems, particularly with OpenAI and Anthropic integrations\n- Plugin order behavior was explained: the order determines which models get registered and used first, affecting default model selection\n- Multiple users reported crashes when running agents locally, often due to missing or improperly configured API keys\n- Two separate implementations of an Akash Chat plugin were shared, leading to some community tension\n\n### Knowledge Management & PDF Processing\n- Users discussed PDF processing issues within the Eliza knowledge pipeline, with some PDFs causing \"Invalid array length\" errors\n- PDF RAG functionality works but requires clean PDFs without unusual formatting, breaking large documents into chunks, and validating structure\n- Jin shared updates about a knowledge repository deployment for aggregating context from various ElizaOS-related sources\n- Setting up multiple agents with their own knowledge bases in v2 requires creating separate knowledge directories and updating agent configs\n\n### Deployment & Hosting\n- Phala Cloud was recommended as a free deployment option for Eliza bots\n- Database options for vector storage were discussed, with Supabase and Neon suggested as PostgreSQL alternatives\n- Windows compatibility issues with pgvector were noted, with Docker suggested as a workaround\n\n## Key Questions & Answers\n\n**Q: Does elizaos/core v0.25.9 support model gpt 4.1?**  \nA: That's not ready yet (answered by Odilitime)\n\n**Q: What is the best hosting platform to deploy the Eliza bot?**  \nA: You can deploy one for free on Phala Cloud https://cloud.phala.network/eliza (answered by Agent Joshua ₱ | TEE)\n\n**Q: Is package \"the-org\" like a demo app basically starting up a few example agents?**  \nA: Correct, It's an out of the box example of popular agent examples that one could use for an organization. eg: Community Manager, DevRel, Project Manager etc. (answered by Nisita)\n\n**Q: How can I fix my agent disliking to answer the same question?**  \nA: You'll need to edit the character definition to include logic that identifies price check requests, always provides latest data, and doesn't consider repeated checks as trolling behavior (answered by 0xbbjoker)\n\n**Q: pgvector seems to have major conflicts with Windows and Unix‐only headers. Can anyone recommend an alternative with PostGresSql?**  \nA: Alternatives you have: Supabase, Neon.tech, or spin postgres in docker (answered by 0xbbjoker)\n\n**Q: Does PDF RAG work decently for any dev here?**  \nA: I've had decent results with PDF RAG in Eliza's knowledge pipeline, but it can be finicky. Key things that helped: 1) clean PDFs without weird formatting 2) breaking large docs into chunks 3) validating the PDF structure before ingestion. (answered by Ruby)\n\n## Community Help & Collaboration\n\n- **0xbbjoker helped Cocaine** with agent crashes when running locally by explaining plugin order behavior and recommending removing local models with \"rm -rf ~/.eliza\" to force using OpenAI API\n  \n- **0xbbjoker assisted Berlin** with plugin loading errors with @elizaos/plugin-openai by suggesting rolling back to v0.25\n\n- **Agent Joshua ₱ | TEE helped Baam25** find a hosting platform for Eliza bot by recommending Phala Cloud for free deployment\n\n- **Odilitime guided standard** on implementing a RabbitMQ-like queue system in v2 by recommending the tasks system and providing documentation links\n\n- **DeFine helped Cocaine** resolve agent crashes by explaining the need to add Anthropic API key to .env file\n\n- **0xbbjoker assisted Hans** with an agent refusing to answer repeated questions by suggesting modifications to character definition components\n\n- **Ruby offered help to mindxploit** with PDF processing issues and setting up multiple agents with their own knowledge bases in v2\n\n## Action Items\n\n### Technical\n- Fix bug where agents default to Anthropic and error when key not specified (Mentioned by DeFine)\n- Resolve plugin compatibility issues between v1 and v2 (Mentioned by Odilitime)\n- Fix Windows compatibility issues with ElizaOS v2 (Mentioned by Cocaine)\n- Check pull request for plugin-bnb (Mentioned by AIFlow.ML @ ElizaOS)\n- Validate PDF structure before ingestion to prevent parsing errors (Mentioned by Ruby)\n- Break large PDF documents into chunks for better processing (Mentioned by Ruby)\n\n### Documentation\n- Create guide for character/personality creation for agents (Mentioned by Smirking_Stoic)\n- Update documentation on plugin order behavior and API key configuration (Mentioned by 0xbbjoker)\n- Improve documentation on setting up multiple agents with their own knowledge bases in v2 (Mentioned by mindxploit)\n\n### Feature\n- Improve Twitter integration for bots (Mentioned by Baam25)\n- Add support for GPT-4.1 in elizaos/core (Mentioned by Sabochee)\n- Implement endless conversations between two AI agents (Mentioned by desu)\n- Enhance PDF parsing error handling (Mentioned by mindxploit)\n- New Akash Chat plugin implementation (Mentioned by Fenil Modi and Carlos Guimaraes)"
        }
      ]
    }
  },
  "github": {
    "summaries": {
      "week": [
        {
          "filename": "2025-04-27.md",
          "content": "# ElizaOS Weekly Update (Week of Apr 27, 2025)\n\n## OVERVIEW\nThis week saw significant improvements to the ElizaOS framework with a focus on enhancing developer experience, fixing critical bugs, and improving plugin functionality. Key achievements include scopable knowledge implementation, Discord plugin enhancements, and comprehensive CLI command updates.\n\n## PROJECT METRICS\n- PRs: 29 merged PR's, 31 new PRs\n- Issues: 0 new issues, 0 closed issues\n- Unique Contributors: 23\n- Code Changes: +6059/-21763 lines across 140 files\n- Total Commits: 92\n- Most Active Contributors: wtfsayo, lalalune, 0xbbjoker, tcm390, samarth30\n\n## TOP ISSUES\n\n### Plugin Integration Challenges\n- Several issues related to plugin functionality were addressed, particularly with Discord, Anthropic, and Twitter integrations\n- Error handling improvements were needed for various edge cases, including disk space limitations\n\n### Developer Experience Friction\n- CLI command instructions required updates to improve clarity and usability\n- Documentation links and image paths needed fixing to ensure proper resource accessibility\n- Environment setup and configuration issues in Docker containers needed resolution\n\n## KEY TECHNICAL DEVELOPMENTS\n\n### Enhanced Plugin Functionality\n- Added typing indicator support in Discord plugin [#4364](https://github.com/elizaos/eliza/pull/4364)\n- Implemented API key validation for Anthropic plugin model calls [#4383](https://github.com/elizaos/eliza/pull/4383)\n- Fixed Discord plugin error handling when services are missing [#4375](https://github.com/elizaos/eliza/pull/4375)\n- Added model/plugin name logging when models are used [#4394](https://github.com/elizaos/eliza/pull/4394)\n\n### Improved Developer Experience\n- Implemented scopable knowledge functionality [#4390](https://github.com/elizaos/eliza/pull/4390)\n- Updated ElizaOS create command instructions [#4381](https://github.com/elizaos/eliza/pull/4381)\n- Added automatic rebuild of core and plugin-bootstrap in monorepo context [#4388](https://github.com/elizaos/eliza/pull/4388)\n- Added .env.example in project-starter for easier configuration [#4387](https://github.com/elizaos/eliza/pull/4387)\n- Fixed ESM type declarations in Core [#4341](https://github.com/elizaos/eliza/pull/4341)\n- Updated documentation with improved quick start and CLI commands [#4379](https://github.com/elizaos/eliza/pull/4379)\n\n### Infrastructure and Maintenance\n- Fixed Postgres database issues in Docker containers [#4363](https://github.com/elizaos/eliza/pull/4363)\n- Improved error handling for disk space limitations [#4389](https://github.com/elizaos/eliza/pull/4389)\n- Fixed agent deletion functionality [#4395](https://github.com/elizaos/eliza/pull/4395)\n- Removed plugin-evm from monorepo [#4386](https://github.com/elizaos/eliza/pull/4386), [#4399](https://github.com/elizaos/eliza/pull/4399)\n- Enhanced UI with cleaner design and minor improvements [#4393](https://github.com/elizaos/eliza/pull/4393)"
        }
      ],
      "month": [
        {
          "filename": "2025-04-01.md",
          "content": "# ElizaOS Monthly Update (April 2025)\n\n## OVERVIEW\nApril saw significant enhancements to ElizaOS with major improvements to the CLI, GUI, and plugin ecosystem. Key achievements include expanded AI model provider integrations, enhanced messaging capabilities, improved TTS functionality, and substantial fixes to core components and plugins.\n\n## PROJECT METRICS\n- PRs: 195 merged PR's, 232 new PRs\n- Issues: 43 new issues, 35 closed issues\n- Unique Contributors: 88\n- Code Changes: +55467/-37251 lines across 489 files\n- Total Commits: 828\n- Most Active Contributors: wtfsayo, tcm390, 0xbbjoker, Samarthsinghal28, HarshModi2005\n\n## TOP ISSUES\n\n### CLI Usability and Consistency\n- Command inconsistencies between v1 and v2\n- Installation and update workflow issues\n- Project creation and agent management challenges\n\n### Plugin Management\n- Plugin installation priority ordering issues\n- Dynamic loading and versioning problems\n- Plugin dependency conflicts\n\n### AI Model Integration\n- JSON parsing errors with Anthropic and OpenAI responses\n- Model validation and authentication issues\n- Embedding model selection and configuration\n\n### Database and Transaction Handling\n- Transaction deadlocks in database operations\n- Migration issues with PgLite\n- Entity relationship inconsistencies\n\n### Messaging and Communication\n- Duplicate message creation\n- Twitter interaction failures\n- Telegram data synchronization problems\n\n### User Interface\n- Knowledge manager scroll issues\n- Missing UI components after merge conflicts\n- Inconsistent character creation workflow\n\n### Performance and Resource Management\n- Disk space management and error handling\n- Memory leaks and caching issues\n- Runtime initialization sequence problems\n\n### Documentation and Onboarding\n- Inconsistent or outdated CLI documentation\n- Missing environment configuration examples\n- Broken documentation links\n\n## KEY TECHNICAL DEVELOPMENTS\n\n### Enhanced AI Model Provider Ecosystem\n- Added Kluster AI as a model provider [#3938](https://github.com/elizaos/eliza/pull/3938)\n- Added Mem0 as AI SDK Provider [#3927](https://github.com/elizaos/eliza/pull/3927)\n- Improved OpenAI plugin with ability to choose embedding models [#4140](https://github.com/elizaos/eliza/pull/4140)\n- Added model/plugin name logging when model is used [#4394](https://github.com/elizaos/eliza/pull/4394)\n\n### Voice and Speech Capabilities\n- Implemented OpenAI TTS functionality [#4255](https://github.com/elizaos/eliza/pull/4255)\n- Added testing for OpenAI TTS [#4259](https://github.com/elizaos/eliza/pull/4259)\n- Fixed speech-to-text transcription issues [#4236](https://github.com/elizaos/eliza/pull/4236)\n- Fixed OpenAI STT functionality [#4329](https://github.com/elizaos/eliza/pull/4329)\n\n### CLI Improvements\n- Added update-cli command for easier updates [#4170](https://github.com/elizaos/eliza/pull/4170)\n- Enhanced CLI update and plugin installation with package manager [#4176](https://github.com/elizaos/eliza/pull/4176)\n- Implemented CLI test suite [#4301](https://github.com/elizaos/eliza/pull/4301)\n- Updated ElizaOS create command instructions [#4381](https://github.com/elizaos/eliza/pull/4381)\n\n### Messaging and Interaction Enhancements\n- Added message API for improved communication [#4247](https://github.com/elizaos/eliza/pull/4247)\n- Added message buttons for interactive responses [#4187](https://github.com/elizaos/eliza/pull/4187)\n- Improved Reply Action to use pre-generated responses [#4213](https://github.com/elizaos/eliza/pull/4213)\n- Added typing indicators to Telegram [#4280](https://github.com/elizaos/eliza/pull/4280) and Discord [#4364](https://github.com/elizaos/eliza/pull/4364)\n\n### Plugin System Improvements\n- Better plugin installation management [#4177](https://github.com/elizaos/eliza/pull/4177)\n- Added test coverage for plugin-bootstrap [#4173](https://github.com/elizaos/eliza/pull/4173)\n- Enabled path params routes for plugins [#4335](https://github.com/elizaos/eliza/pull/4335)\n- Fixed plugin install priority order [#4256](https://github.com/elizaos/eliza/pull/4256)\n\n### GUI Enhancements\n- Added support for importing JSON to create or update agents [#4270](https://github.com/elizaos/eliza/pull/4270)\n- Enhanced GUI with various improvements [#4300](https://github.com/elizaos/eliza/pull/4300)\n- Added confirmation dialog for delete group [#4310](https://github.com/elizaos/eliza/pull/4310)\n- Implemented onboarding tour [#4293](https://github.com/elizaos/eliza/pull/4293)\n\n### Social Media Integration\n- Enhanced Telegram logic with middlewares [#4125](https://github.com/elizaos/eliza/pull/4125)\n- Added Telegram community manager [#4134](https://github.com/elizaos/eliza/pull/4134)\n- Improved Twitter plugin with support for long tweets [#4291](https://github.com/elizaos/eliza/pull/4291)\n- Added deleteTweet() functionality [#4320](https://github.com/elizaos/eliza/pull/4320)\n\n### Knowledge and Memory Management\n- Implemented scopable knowledge [#4390](https://github.com/elizaos/eliza/pull/4390)\n- Fixed knowledge types [#4243](https://github.com/elizaos/eliza/pull/4243)\n- Replaced fastembed with local embedding model [#4205](https://github.com/elizaos/eliza/pull/4205)\n- Fixed cache interaction cursor and duplicate memory creation [#4155](https://github.com/elizaos/eliza/pull/4155)\n\n### Logging and Monitoring\n- Added OpenAI model usage events [#4230](https://github.com/elizaos/eliza/pull/4230)\n- Restored support for Cloudflare LLM logging/tracking [#4354](https://github.com/elizaos/eliza/pull/4354)\n- Added docstrings to cloudflare-logging [#4355](https://github.com/elizaos/eliza/pull/4355)\n- Improved logging for transcription models [#4283](https://github.com/elizaos/eliza/pull/4283)\n\n## SUMMARY\nApril 2025 was a productive month for ElizaOS with significant improvements across the platform. The team expanded AI model provider integrations, enhanced messaging capabilities, improved voice and speech features, and fixed numerous issues in the core system and plugins. The CLI and GUI received substantial upgrades, making the platform more user-friendly and robust. These developments have strengthened ElizaOS's position as a versatile, modular AI agent framework capable of operating across diverse ecosystems."
        }
      ]
    }
  }
}