{
  "server": "Hyperfy",
  "title": "Hyperfy Discord - 2025-02-22",
  "date": 1740182400,
  "stats": {
    "totalMessages": 651,
    "totalUsers": 46
  },
  "categories": [
    {
      "channelId": "994775534733115412",
      "channelName": "💻│developers",
      "summary": "# Analysis of 💻│developers Discord Chat\n\n## 1. Summary\nThe discussion primarily focused on technical improvements to the Hyperfy platform. A significant change was proposed by Ashxn regarding effects management, moving from `player.setEffect()` to `world.setEffect()` and later to `control.setEffect()` to better track app lifecycles and prevent players from getting stuck with effects. The new effects system includes a \"turn\" option that makes players face the camera direction, useful for weapons and interactive objects.\n\nSeveral community members showcased their projects: a throwing mechanic using the new effects system, an NPC controller with collision detection and gravity, a documentation search tool, and a backend connection for board games. There were discussions about mobile support limitations, VRM avatar handling, and permission systems for collaborative building.\n\nKey technical implementations included hidden props for sensitive data, raycasting for object avoidance, and particle systems. The community also discussed migrating v1 apps to v2, with Saori categorizing apps into those requiring core changes versus those that can be remade without changes.\n\n## 2. FAQ\nQ: How can we prevent players from getting stuck with effects when an app is modified? (asked by Ashxn) A: Move from player.setEffect() to world.setEffect() to track app lifecycle and cancel effects when apps are deleted/modified (answered by Ashxn)\nQ: How do you get rid of the grass in Hyperworld? (asked by MetaRick) A: Currently no no-code way to turn it off; in v1 the grid was just an app you could delete (answered by Ashxn)\nQ: Is there a callback when an animation ends to time projectile spawning? (asked by Shiffty) A: Unanswered\nQ: How would you limit builders to certain areas in a world? (asked by devilsadvocate.sol) A: It's technically challenging as scripts can move meshes anywhere, and there's no way to prevent objects from leaving designated zones (answered by Ashxn)\nQ: Will instanced assets be able to be scaled individually? (asked by TheMattEmpire) A: Unanswered\nQ: What's controlling the draggable aspect of panes? (asked by Omka) A: usePane sets up listeners on the pane header and moves it around (answered by Ashxn)\nQ: How can we download VRMs instead of HYPs? (asked by devilsadvocate.sol) A: You can get the model URL in a script with app.avatar.src and make a download button (answered by Saori)\nQ: How do you hide an avatar from a script? (asked by devilsadvocate.sol) A: Use avatar.active = false to completely unmount nodes (answered by Ashxn)\n\n## 3. Help Interactions\nHelper: Ashxn | Helpee: Shiffty | Context: Implementing throwing mechanics with new effects system | Resolution: Ashxn added a \"turn\" option to effects that makes players face camera direction, helping with throwing mechanics\nHelper: Saori | Helpee: devilsadvocate.sol | Context: Downloading VRM files from avatars | Resolution: Shared code snippet showing how to access avatar source URL with app.avatar.src\nHelper: Ashxn | Helpee: devilsadvocate.sol | Context: Hiding avatar from script | Resolution: Provided code to use avatar.active = false to unmount nodes\nHelper: Ashxn | Helpee: Omka | Context: Understanding pane dragging functionality | Resolution: Explained that usePane sets up listeners on pane headers for movement\nHelper: Ashxn | Helpee: Community | Context: NPC movement with collision | Resolution: Shared a demo of an NPC controller that can traverse colliders and has gravity\nHelper: Saori | Helpee: drdoge.eth | Context: Converting Mixamo animations to Hyperfy format | Resolution: Shared Blender scripts to batch convert Mixamo FBX files to Hyperfy GLB format\n\n## 4. Action Items\nType: Technical | Description: Move setEffect to control system for better app lifecycle management | Mentioned By: Ashxn\nType: Technical | Description: Add a toggle for disabling the default ground/grid | Mentioned By: MetaRick\nType: Technical | Description: Implement hidden props for sensitive data | Mentioned By: devilsadvocate.sol\nType: Technical | Description: Add jump button and interaction equivalent for mobile | Mentioned By: Ashxn\nType: Technical | Description: Fix chat functionality on mobile | Mentioned By: devilsadvocate.sol\nType: Documentation | Description: Create documentation search tool with JSON data | Mentioned By: Omka\nType: Feature | Description: Create permission system with different admin levels | Mentioned By: devilsadvocate.sol\nType: Feature | Description: Add ability to download VRM files directly | Mentioned By: devilsadvocate.sol\nType: Feature | Description: Implement secrets storage in world settings | Mentioned By: devilsadvocate.sol\nType: Technical | Description: Migrate v1 apps to v2 (categorized into those needing core changes vs. not) | Mentioned By: Saori",
      "messageCount": 266,
      "userCount": 16
    },
    {
      "channelId": "1330373197203505185",
      "channelName": "🤖│agents",
      "summary": "# Discord Chat Analysis for \"🤖│agents\" Channel\n\n## 1. Summary:\nThe discussion primarily focused on fine-tuning LLMs versus using context for AI agents. Participants debated the trade-offs between these approaches, with cost efficiency and accuracy being key considerations. They explored when fine-tuning might be preferable (for static baseline behaviors) versus when context is better (for dynamic information). The conversation included analogies to human brain development and learning. There was also brief discussion about implementing AI agents using Hugging Face's API, with one user sharing progress on getting a Mistral model working. Another user mentioned developing a Python-based agent with API integration. The chat included sharing of educational resources about RAG vs. fine-tuning approaches and Microsoft's Muse AI model for game design.\n\n## 2. FAQ:\nQ: What's the difference between fine-tuning and using context? (asked by devilsadvocate.sol) A: Fine-tuning changes the weights on the model to better suit specific needs, while context is information fed to the model at runtime (answered by maximus)\nQ: When would you want to tune a model vs feed it context? (asked by devilsadvocate.sol) A: It's a balance - fine-tuning is good for baseline behaviors while context works better for dynamic information; efficiency is a big factor (answered by maximus)\nQ: How is Hugging Face being used with the Mistral model? (asked by maximus) A: Using Hugging Face with an API key to handle responses and guidance (answered by drdoge.eth)\n\n## 3. Help Interactions:\nHelper: maximus | Helpee: devilsadvocate.sol | Context: Understanding fine-tuning vs context | Resolution: Shared educational resources including articles from DataCamp and Medium\nHelper: maximus | Helpee: drdoge.eth | Context: Understanding Hugging Face implementation | Resolution: Shared information about using Hugging Face API for hosted models\nHelper: devilsadvocate.sol | Helpee: unnamed friend | Context: Setting up an agent for the first time | Resolution: Explained that agents can be programmed by simply telling them what to do\n\n## 4. Action Items:\nTechnical: Implement Python-based agent with API integration | Description: Alternative to JavaScript-based agents that takes an Eliza character file | Mentioned By: devilsadvocate.sol\nTechnical: Test performance differences between tuned models and context-fed models | Description: Compare accuracy and efficiency of both approaches | Mentioned By: devilsadvocate.sol\nFeature: Explore fine-tuning as a service (FTAAS) | Description: Investigate potential for fine-tuning as a service in 2026 | Mentioned By: MetaMike\nDocumentation: Create explanatory materials about fine-tuning vs context | Description: Help average users understand the difference between these approaches | Mentioned By: devilsadvocate.sol",
      "messageCount": 60,
      "userCount": 5
    },
    {
      "channelId": "958209074045026327",
      "channelName": "⚡│general",
      "summary": "# Discord Chat Analysis\n\n## 1. Summary\nThe chat primarily revolves around Hyperfy's development ecosystem, particularly the transition from V1 to V2. Key technical discussions include documentation efforts with mentions of Starlight Astro for documentation hosting, with jin sharing v1 docs on GitHub and Omka working on consolidating documentation from various sources. There's discussion about PhysX support in NVIDIA RTX 50-series GPUs, with clarification that Hyperfy runs physics on CPU with WASM supporting SIMD. Community members are working on in-world documentation and UI components including a .hyp or commands search node. The chat confirms that V1 apps will eventually be ported to V2 and made available on marketplaces like hyperworld.host, though this will be partly community-driven. There's also mention of hyperfy.how, a documentation site created by a community member. The conversation includes references to community projects, animated content created for Hyperfy, and various community events.\n\n## 2. FAQ\nQ: Who made hyperfy.how? (asked by jin) A: Ashxn responded that user 783416470205497394 did it.\nQ: Are there any competitors to Hyperfy at the moment? (asked by cloudAI) A: MetaMike answered that game engines could be considered competitors but the category is too broad.\nQ: Do we still need an nft to create a world? (asked by IamKR) A: Omka answered that for v2, no NFT is needed and provided links to setup instructions.\nQ: When hyper v2 released? (asked by mkl) A: HPrivakos responded \"Over a month ago\".\nQ: Would NVIDIA RTX 50-series not supporting PhysX impact Hyperfy? (asked by HPrivakos) A: Ashxn clarified they're running physics on CPU, and their WASM supports SIMD.\nQ: Would a server hosting hyperfy benefit from having a GPU that supports PhysX? (asked by HPrivakos) A: Unanswered directly, but related to Ashxn's response about CPU usage.\nQ: I still don't understand how physx works in the browser without a Nvidia card (asked by devilsadvocate.sol) A: Ashxn explained it's just a physics engine that simulates based on colliders.\nQ: Will the apps from V1 be ported over to V2 at some stage? (asked by Paradoxx) A: Saori confirmed \"yes\".\nQ: I change/edit animation_url to glb_url after i mint the 3D asset, right? (asked by treed) A: Unanswered\n\n## 3. Help Interactions\nHelper: Omka | Helpee: IamKR | Context: Asked if NFT is still needed to create a world in V2 | Resolution: Omka explained that NFTs are no longer required and provided links to setup instructions.\nHelper: Ashxn | Helpee: devilsadvocate.sol | Context: Confusion about how PhysX works in browser without Nvidia card | Resolution: Explained it's just a physics engine that simulates based on colliders.\nHelper: devilsadvocate.sol | Helpee: Paradoxx | Context: Needed link to public V2 play space | Resolution: Provided \"Play.hyperfy.xyz\" link.\nHelper: Saori | Helpee: Paradoxx | Context: Suggestion to add V2 play space link to a specific channel | Resolution: Saori confirmed it was added.\n\n## 4. Action Items\nTechnical: Consolidate documentation from various sources into hyperfy.how | Description: Omka working on moving content from various docs/references | Mentioned By: Omka\nTechnical: Create in-world documentation | Description: Working on putting all docs in world with UI components | Mentioned By: Omka\nTechnical: Develop .hyp or commands search node for in-world reference | Description: For easier documentation access in-world | Mentioned By: Omka\nFeature: Port V1 apps to V2 | Description: Make V1 apps available on V2 marketplaces like hyperworld.host | Mentioned By: Saori\nDocumentation: Add each app that can be remade from v1 to GitHub board | Description: Tracking which V1 apps can be ported to V2 | Mentioned By: Saori\nDocumentation: Avoid using Mona's content in documentation | Description: TheMattEmpire noted concerns about using content from previous employer | Mentioned By: TheMattEmpire",
      "messageCount": 86,
      "userCount": 23
    },
    {
      "channelId": "1326789867312775290",
      "channelName": "🪙│hyper",
      "summary": "# Discord Chat Analysis for Channel \"🪙│hyper\"\n\n## 1. Summary:\nThis chat segment contains minimal technical discussion. Participants primarily discussed cryptocurrency market sentiment around the HYPER token, with several price checks using the Rick bot. There was brief nostalgia about Ethereum mining and a mention of potential future \"mining\" within Hyperfy game loops. The conversation included comments about DCA (Dollar Cost Averaging) strategies during market recovery and aspirations for Hyperfy to become \"the next uniswap.\" No significant technical problems were solved or implementations discussed in this segment.\n\n## 2. FAQ:\nQ: When hyper v2 released? (asked by mkl) A: like 2 months ago ish (answered by devilsadvocate.sol)\nQ: Is this bearish or bullish (asked by coard) A: Unanswered\nQ: rick can give things? (asked by Omka) A: Unanswered\n\n## 3. Help Interactions:\nHelper: devilsadvocate.sol | Helpee: mkl | Context: Question about when Hyper v2 was released | Resolution: Informed that it was released approximately 2 months ago\n\n## 4. Action Items:\nFeature: Token flywheel implementation when market recovers | Description: Implementing a token flywheel mechanism for when the market recovers | Mentioned By: coard",
      "messageCount": 29,
      "userCount": 11
    },
    {
      "channelId": "1031058655581323324",
      "channelName": "🧊│3d-design",
      "summary": "# Discord Chat Analysis: 🧊│3d-design\n\n## 1. Summary\nThe discussion primarily focused on optimization techniques for 3D assets in web-based environments like Hyperfy. TheMattEmpire proposed a material batching approach where materials could be applied within Hyperfy rather than embedded in GLB files, potentially enabling custom shaders. The conversation evolved into detailed discussions about optimization constraints in WebGL versus native applications, with particular emphasis on avatar file sizes and polygon counts. \n\nSeveral optimization techniques were highlighted, including palette texturing with small (16x16 or 32x32) textures to create lightweight assets. TheMattEmpire shared examples of characters created using this method. The group discussed the technical limitations of browsers compared to native applications, with WebGL having hard limits that WebGPU might eventually overcome.\n\nThe conversation touched on balancing visual quality with performance, noting that conceptual design and storytelling can create immersive experiences without requiring high-polygon models. Participants acknowledged that while triple-A games use avatars with approximately 80k polygons, web-based environments require much more optimization (suggested 15k polygons or fewer for avatars). A key concern was the cumulative resource load when multiple users with heavy avatars join a space.\n\n## 2. FAQ\nQ: How can materials be applied to GLB files in Hyperfy? (asked by TheMattEmpire) A: Unanswered\nQ: Is the Blender exporter documentation still necessary to show? (asked by Omka) A: It's recommended not to delete it as it gives people who don't want to download Hyperfy Tools an option (answered by maximus)\nQ: What's the ideal polygon count for web avatars? (asked by devilsadvocate.sol) A: Around 15k is medium, but it depends on your end experience and must consider material count and texture resolution (answered by TheMattEmpire)\nQ: What limits web performance more - polygons or textures? (asked by devilsadvocate.sol) A: For filesize, textures are worse than polygons. For processing cost, it depends (answered by TheMattEmpire)\n\n## 3. Help Interactions\nHelper: TheMattEmpire | Helpee: devilsadvocate.sol | Context: Understanding optimization requirements for web-based 3D assets | Resolution: Explained that web browsers have hard limits unlike native clients, and that optimization depends on the intended experience\nHelper: Ashxn | Helpee: Multiple users | Context: Explaining why avatar optimization matters | Resolution: Clarified that while performance might be fine on good desktops, downloading many large avatars (e.g., 100 people with 30MB avatars = 3GB) creates significant problems\nHelper: ToxSam | Helpee: devilsadvocate.sol | Context: Error when downloading avatars from opensourceavatars.com | Resolution: Acknowledged the issue and committed to fixing it the following week\n\n## 4. Action Items\nType: Technical | Description: Explore material batching approach where materials are applied within Hyperfy rather than embedded in GLB files | Mentioned By: TheMattEmpire\nType: Technical | Description: Fix download error on opensourceavatars.com website | Mentioned By: ToxSam\nType: Documentation | Description: Decide whether to keep Blender exporter documentation or direct users to Hyperfy Tools | Mentioned By: Omka\nType: Feature | Description: Create a rigging course (15 minutes max) | Mentioned By: ᲼\nType: Feature | Description: Create a \"make your first animation for hyperfy avatars\" course | Mentioned By: peezy",
      "messageCount": 105,
      "userCount": 13
    },
    {
      "channelId": "999870741069692958",
      "channelName": "🌆│share",
      "summary": "# Analysis of Discord Channel \"🌆│share\"\n\n## 1. Summary\nThe chat segment is very brief, containing only two messages. There are no technical discussions, problem-solving, or implementations shared. The first message from Rick shares an animated short inspired by Hyperfy metaverse created by user \"scethmedia\" with a link to a tweet. The second message from VTATV promotes a live music event at a virtual New York nightclub in the Hyperfy platform, scheduled for 10pm EST, with links to both hyperfy.io/vtatv and vtatv.io.\n\n## 2. FAQ\nNo questions were asked in this chat segment.\n\n## 3. Help Interactions\nNo help interactions occurred in this chat segment.\n\n## 4. Action Items\nNo action items were identified in this chat segment.",
      "messageCount": 4,
      "userCount": 3
    },
    {
      "channelId": "1022421054582423562",
      "channelName": "🐞│issues",
      "summary": "# Discord Chat Analysis for 🐞│issues Channel\n\n## 1. Summary:\nThe discussion primarily revolves around the `world-clean.mjs` script (also accessible via `npm run world:clean`), which removes unnecessary files from Hyperfy worlds. Users share experiences running this script manually at different intervals. A key technical insight from ash is that the cleaner should not be run while a world is active as it could delete files not yet saved to the database. HypPrivakos reports that the script sometimes \"kills\" worlds exceeding 1GB that haven't been cleaned in over a month. A GitHub issue (#50) was reported by general hyper and added to the project board by Saori, related to problems that occur when cleaning without restarting the server. The conversation also briefly touches on Coolify as a deployment solution, with HowieDuhzit mentioning a docker-compose setup for deploying modded Hyperfy instances. Ash suggests adding a command to force saves, noting that users have requested both manual save functionality and the ability to disable auto-save.\n\n## 2. FAQ:\nQ: What does `node world-clean.mjs` do? (asked by general hyper) A: It cleans all the unnecessary leftovers from 'deleting'. (answered by 0mka)\nQ: Should world-clean run during server startup? (asked by ash) A: It might be risky as it could potentially delete the wrong things. (answered by ash)\nQ: Can you push servers to the latest? (asked by vox) A: Yes, and a main branch release with all the new features was just made. (answered by ash)\nQ: Does world:clean have any performance impact, or is it really just to clean the world folder? (asked by HypPrivakos - HyperWorld.Host) A: It seemed to impact Doge's world but wasn't thoroughly tested; definitely keeps folder size down. (answered by HowieDuhzit)\nQ: Is there a way to force a save with a command like the world clean? (asked by HowieDuhzit) A: Not currently, but could be added. Users have also requested disabling auto-save for a manual save button/command. (answered by ash)\n\n## 3. Help Interactions:\nHelper: 0mka | Helpee: general hyper | Context: Explaining what world-clean.mjs does and how to run it | Resolution: Provided command and explanation of its purpose\nHelper: ash | Helpee: general hyper | Context: Providing the npm command for world cleaning | Resolution: Shared the proper npm command: \"npm run world:clean\"\nHelper: Saori | Helpee: general hyper | Context: Adding GitHub issue to project board | Resolution: Successfully added issue #50 to the project board\nHelper: HypPrivakos | Helpee: general hyper | Context: Explaining GitHub issue #50 about page loading failures | Resolution: Identified it happens when running world clean without restarting the server\nHelper: ash | Helpee: Multiple users | Context: Explaining dangers of running world-clean while world is active | Resolution: Detailed explanation of how it could delete files not yet saved to database\n\n## 4. Action Items:\nTechnical: Don't run world-clean.mjs while a world is running to prevent data loss | Mentioned By: ash\nTechnical: Debug why world-clean kills some hosted worlds exceeding 1GB | Mentioned By: HypPrivakos - HyperWorld.Host\nTechnical: Implement a force save command for worlds | Mentioned By: ash\nTechnical: Create a save/shutdown/clean/restart cycle for safer world cleaning | Mentioned By: HowieDuhzit\nFeature: Add option to disable auto-save and implement a manual save button/command | Mentioned By: ash\nDocumentation: Create a tutorial on using Coolify for Hyperfy deployment | Mentioned By: ash\nDocumentation: Document the proper workflow for running world-clean safely | Mentioned By: ash",
      "messageCount": 48,
      "userCount": 9
    },
    {
      "channelId": "1332108186676891649",
      "channelName": "🏗│infra",
      "summary": "# Analysis of 🏗│infra Discord Channel\n\n## 1. Summary\nThe discussion primarily revolves around hosting solutions for HyperWorld instances. Users discussed pricing models, with estimates ranging from $3.2-9 per world per month, with potential bulk discounts. HypPrivakos mentioned running a hosting service (hyperworld.host) with a $5 USD plan. \n\nA significant technical issue was identified by ash regarding websocket URL configuration in the .env file, where http:// was incorrectly used instead of ws:// or wss://, causing loading problems specifically on iOS devices and Safari while working fine on Chrome. This highlights a cross-browser compatibility issue.\n\nAnother technical problem discussed was related to a \"world-clean\" feature that was deleting necessary files, causing 404 errors. HypPrivakos restored these files from a backup to resolve the issue for a specific world (vox.hyperworld.host).\n\nThere was also a brief discussion about the infrastructure stack, confirming that Cloudflare R2 is S3-compatible, and questioning whether the main server would use Docker containers.\n\n## 2. FAQ\nQ: What's a fair price for per-world hosting? (asked by DefiniteOptimist) A: $9/mo sounds fair, could reduce to $5 through bulk rates (answered by DefiniteOptimist)\nQ: Is R2 compatible with S3? (asked by smickelbeard) A: Yes, R2 is S3-compatible (answered by HypPrivakos - HyperWorld.Host)\nQ: Would the main server spin up docker containers or something? (asked by smickelbeard) A: Unanswered\n\n## 3. Help Interactions\nHelper: ash | Helpee: DevStarlight and others | Context: Incorrect websocket URL protocol in .env.example (http:// instead of ws://) causing loading issues on iOS devices | Resolution: Advised to check and ensure ws:// or wss:// is used\nHelper: HypPrivakos - HyperWorld.Host | Helpee: ash | Context: Vox's world not loading on iPhone | Resolution: Restored cleaned files from backup that were erroneously deleted, fixing the loading issue\n\n## 4. Action Items\nTechnical: Fix the .env.example file to use ws:// or wss:// instead of http:// for websocket URLs | Mentioned By: ash\nTechnical: Investigate why the world-clean feature is deleting necessary files | Mentioned By: ash\nDocumentation: Update documentation to specify the correct websocket protocol (ws:// or wss://) for different browsers | Mentioned By: ash",
      "messageCount": 18,
      "userCount": 5
    },
    {
      "channelId": "1332430296737644677",
      "channelName": "🎨│showcase",
      "summary": "# Analysis of 🎨│showcase Channel\n\n## 1. Summary\nThe chat primarily revolves around racing game development in Hyperfy, with TheMattEmpire working on road components for a racing kit. The discussion includes various road elements like ramps, slopes, and bridges that would support racing gameplay. TheMattEmpire is planning to separate features into different kits (roads, props, boost/powerups) for modularity. There's consideration about how these elements will work with future vehicle implementations, particularly regarding physics and \"feel.\"\n\nIn parallel, ~/MayD524/Hyperfy is developing a pet companion app that follows players, avoids obstacles, teleports when stuck, and is fully networked. The developer is working on reducing the model size and fixing bugs before release.\n\nTechnical considerations mentioned include VR compatibility (requiring 90 FPS minimum in browser), physics for racing elements like loops and banked turns, and networking for multiplayer synchronization of pets.\n\n## 2. FAQ\nQ: Do you plan to release this as a \"pets\" app? (asked by Saori) A: Yes I do plan to release this as a \"pets\" app, right now I'm fixing a few bugs and getting a different model this one is like 17mb which doesn't fly with me (answered by ~/MayD524/Hyperfy)\nQ: Can I buy a basic kit now? Pretty much just want straights, curves, and a 4 way intersection (asked by general hyper) A: Unanswered\nQ: How is minting a VRM avatar different from minting a 3D asset on manifold? (asked by treed) A: Unanswered\n\n## 3. Help Interactions\nHelper: TheMattEmpire | Helpee: 0mka | Context: Discussion about what road elements to include in racing kit | Resolution: TheMattEmpire confirmed adding up/down slopes to the list and shared progress on ramps\nHelper: ~/MayD524/Hyperfy | Helpee: Saori | Context: Question about releasing pet companion as an app | Resolution: Developer confirmed plans to release it as an app after fixing bugs and reducing model size\n\n## 4. Action Items\nType: Technical | Description: Reduce pet companion model size from 17mb | Mentioned By: ~/MayD524/Hyperfy\nType: Technical | Description: Fix bugs in pet companion app before release | Mentioned By: ~/MayD524/Hyperfy\nType: Feature | Description: Add bridges to the road kit | Mentioned By: TheMattEmpire\nType: Feature | Description: Create separate kit for boost and power ups | Mentioned By: TheMattEmpire\nType: Feature | Description: Add pit road elements to racing kit | Mentioned By: TheMattEmpire\nType: Feature | Description: Consider adding over-banked curves to road kit | Mentioned By: Agent12\nType: Technical | Description: Ensure VR racing compatibility at 90 FPS minimum | Mentioned By: TheMattEmpire",
      "messageCount": 35,
      "userCount": 9
    }
  ]
}