# 1,600 stars, 0 downloads: an open-source library that turns AI Dungeon into video, and its fork every 20 minutes · GitHub open-source project > Source: https://stringers.io/en/github/gh_Murea-Labs__InWorld.html (Stringers / 什么值得做, generated 2026-11-08) > Stringers answers one question every day: what small product can one or two people build in a few months and actually make money from. Numbers come from 15 databases we collect daily, each with source and date. > For agents: these verdicts and the underlying databases (verified revenue, App Store / Google Play charts, ad libraries, GitHub, Product Hunt…) are queryable live over MCP — tools today / hot / products / directions / direction / data. Setup: https://github.com/komako-workshop/stringers (the user logs in at stringers.io → "配置 agent" to get a personal URL). Site index: https://stringers.io/llms.txt > It never really took off: a set of Machine Heart articles gave it a bit of exposure, and the stars and forks after that were bought at a steady pace. A fork arrives every 20 minutes, all from zero-follower alt accounts, while watchers, issues, and Hugging Face downloads are 0. [GitHub open-source project](https://stringers.io/en/github/index.md) · written 11-6 Murea-Labs/InWorld · Murea-Labs # 1,600 stars, 0 downloads: an open-source library that turns AI Dungeon into video, and its fork every 20 minutes **What it is** An open-source interactive video world: an agent remembers characters, items, and plot progress, and a video model films each action you type as a short clip with sound, with rollback and multiplayer support. **Why it blew up** It didn't really blow up: a series of articles from Synced gave it a bit of exposure, and the stars and forks since then were added at a steady rate. Forks arrive about every 20 minutes, all from zero-follower alt accounts, while watchers, issues, and Hugging Face download counts are 0. **D** **How you could approach it** Don't build a world model, just build its agent layer: an open-source "interactive story director" that uses a ledger and reference images to keep characters consistent, hands the visuals to an existing video API, runs without a GPU, and puts a dozen-plus rounds of continuous plot and a live demo on the README's first screen. Grade D: its stars were bought, which doesn't prove anyone wants this; on the model layer the competitors are Ant, Tencent, and Google, which a small team can't match. What's worth learning is the cold-start packaging and the "agent manages world state" structure: worth understanding, not recommended to copy as-is. Spreading in small circles Data (updated daily) · Repo created 2026-10-13 · 1,849 stars · 7 days +954 GitHub · stars Data 2026-10-21 → 2026-11-08 (19 days, 18 observations). Points before 10-28 come from deeply / archive backfill, not every day has one Murea-InWorld is an open-source interactive audio-visual world model: you type "what happens next," and it films the story as a continuous short clip with sound. In 24 days it went from 25 stars to over 1,600, which looks like a project gaining momentum. Taken apart, the real exposure is one set of articles from a Chinese AI outlet, and the rest of the growth has a clearly machine-like rhythm. Below: what it is, how the numbers came about, and whether this direction is worth building for you. ### Five forks from the evening of Nov 6 On Nov 6 of 2026, the five newest forks of InWorld were created at Beijing time 16:44, 17:26, 18:05, 18:45, and 19:05, by accounts named ayasma37, zerocearlo, jjohnnettenduara, yxemg, and atrothasantomaim, and the forked repos were renamed hzjlwiaow, tawafqclqgx, rnpujnksje, qcireoghnj, and jujtxcbncq (GitHub, 11-06). This isn't a coincidence. Looking back, 259 forks in total, 229 of them were created just after the top of the hour, at :02, :22 or :42, within a few seconds of each other. Starting Oct 25 00:02:09, that's 19 to 23 a day, without a single day off. The accounts I spot-checked each have 0 followers, 0 following, and 1 to 5 repos. The same accounts also forked other repos, like one project with 23 stars and 251 forks. This is a pool of accounts shared by several parties. ### What it is Murea Labs is a GitHub org created in Oct 2026 12, with just this one repo. Judging from the Chinese write-ups and partners, it's most likely a China-based team (inference). InWorld has two parts: - **Murea-Mind** : a "world state" agent that acts as screenwriter and director. It keeps track of characters, items, locations, and plot progress, reads your input and the last frame of the previous clip, plans the next event, and after generation reviews the footage and writes the actual changes back to the "ledger." It calls Gemini, OpenAI-compatible APIs, or Alibaba's DashScope underneath. - **Murea-Fast** : a generation service built on Lightricks' open-source LTX video model, where characters, cats, props, and scenes each get a reference image to lock their look, and the video continues segment by segment, with picture and sound produced together. It supports rollback and branching, and 2 to 4 people can co-write a story in one room. The official site's first screen is a 2-minute 19-second demo video with the line "Create a world, choose your next move, and watch the story unfold with video and sound." In short, it "films" the kind of text adventure found in AI Dungeon. Two things to watch: the top-scoring flagship Murea-Pro is not open source; what's open is the lightweight Murea-Fast. And according to the docs, Fast's weights need 4 H20 GPUs to run, and there is no online demo on the web. ### Where the stars came from **A set of media articles, with real exposure, but it didn't move the numbers.** On Oct 18, Synced published "Top of Two Leaderboards: What They Want Is Not the Next World Model," covering how Murea-Pro scored 81.1 to take first place on Meituan LongCat's interactive video benchmark WBench, and how Murea-Fast ranked first among open-source models on OpenS2V-Eval. Tencent News and Sohu reposted it. On Oct 25 a video followed (the YouTube version had 92 views), and on Oct 27 Synced's X account posted an English tweet tagged "#PapersAccepted," the name of its paper-submission column. The articles read throughout like promotional copy, like a team submission or a sponsored piece (whether it was paid is unverified). In the week after publication, stars only rose from about 25 to 29. **Then the curve suddenly flattened.** From Oct 25 the count went from 29 stars, with 26 on 163, and afterward gained +77 to +153 a day, most days holding steady between 120 and 135 (GitHub daily data, 10-21 to 11-06). On Nov 6 it was 1,559 stars, and a live check that afternoon showed 1,693 stars and 251 forks. Normal media exposure spikes and then falls off; it doesn't stay at nearly the same daily amount for over ten days. **Three numbers that don't add up.** Behind the 1,600+ stars: watchers are 0, open issues are 0, and the weight downloads on Hugging Face are 0 (Hugging Face, 11-06). The last code commit was on Oct 20, and the repo hasn't changed since stars started climbing. In other words, almost nobody actually downloaded, ran, or raised issues with it. **Almost nobody knows about it overseas.** Searches on X, Reddit, and Hacker News turn up no discussion; secondary spread is limited to a few Chinese aggregator sites and one personal notes library. Conclusion: **this is a sample of "one media push plus bought forks / bought stars," not organic growth.** Whether the stars were all bought can't be checked one by one (GitHub no longer provides per-star timestamps), but the fork evidence combined with the steady curve makes it highly likely (inferred). Who bought the traffic can't be determined from the outside. ### Whether the direction itself is interesting Yes. The hardest part of long-form story video is continuity: by the third segment the cat has changed color and the lead is wearing different clothes. InWorld doesn't try to make the video model remember this. It puts an agent outside the model to keep a "ledger": fixed identity (reference images, voice) is managed separately from state that changes (position, mood, what's in hand), and every segment gets reviewed after it's generated. The same approach works for interactive short dramas and AI comic series. But the competition on this path is heavy: | Project | Form | Stars or scale | How it differs from InWorld | |---|---|---|---| | Ant LingBot-World | Open source, 2026-01 | 4,571 star | Real-time, controllable world exploration; no story state or dialogue voice | | Kunlun Matrix-Game | Open source, 2025-05 | 2,855 star | Keyboard/mouse-controlled game-style world; no text storytelling | | Tencent HunyuanWorld / GameCraft | Open source, 2025-07 / 08 | 2,961 / 767 star | 3D scene or game video | | Lightricks LTX-2 | Open source, 2026-01 | 10,101 star | The base InWorld builds on; general audio-video generation | | Alibaba HappyOyster, Shengshu Vidu S2, Aishi PixVerse R2 | Commercial products | Has real users | Closed-source interactive video, with built-in distribution | | Google Genie 3, World Labs Marble | big tech | Industry benchmarks | Explorable worlds, not focused on story agents | | AI Dungeon and similar | Web / App, starting at 2019 | Millions of users | Text only; InWorld wants to turn it into video | Building the "world model" itself means training models, stacking compute, and chasing evaluation leaderboards. Your opponents are Ant, Tencent, and Google. An individual or small team can't win stars at this layer, let alone real users. ### What to learn, and what not to **Worth learning: the cold-start packaging of an AI lab.** A technical report PDF, a polished project page with a demo video, a bilingual README, Hugging Face weights, and a feature in a top media outlet. Put together, it looks like a legitimate lab. Its website and README show 15 commits credited to "Codex", so this packaging was fast to produce. **Worth learning: the "agent keeps the ledger" structure.** This layer doesn't require training a model. You can build it on off-the-shelf video APIs. **Don't copy: buying stars.** Cross-check watchers, download counts, and fork timestamps and the pattern gives it away. Bought numbers also buy you nothing: no one downloads, no one opens issues, no one reposts. The project went quiet in Oct 20. ### If it were you **Bottom line:** Don't build an "open-source world model" after this one, and don't treat its stars as evidence of demand. The only part worth thinking about is the part it did badly: getting ordinary people to actually play it. **Who it's for:** Indie developers comfortable with agents and video APIs who want an open-source project that can earn stars; small teams making interactive short dramas or AI comics who want an open "story consistency" layer. **Who it's not for:** Anyone who wants to train their own video model and climb evaluation leaderboards (your opponents are big tech); anyone expecting 1,600 real users waiting for it (they don't exist). **How to do it:** Build only the Murea-Mind layer: an open-source "interactive story director" agent with a world-state ledger, reference images to lock characters, rollback branches, and multiplayer rooms. Leave the visuals to existing video APIs (Veo, Kling, hosted LTX), so it runs without a single GPU. Put a dozen-plus turns of continuous story that plays directly at the top of the README, plus a live demo link. InWorld lacks exactly these two things: something that runs, and something people can play. **How to get users:** Package it as a skill or MCP that Claude Code / Codex can install directly, and get it into awesome lists and skill directories. On English X, post a side-by-side clip of "the same cat staying unchanged across ten story turns": consistency is the easiest and most visual selling point in this category. In the Chinese community, pitch newsletters for features. **When it's too early:** Interactive video has been a direction every major player has been shipping for 2026 years. Once a big vendor builds "world state" into its API, this layer gets eaten. Shipping within three months is fairly safe (my estimate). **Signs you've misjudged it:** A month after launch, no one has played the live demo for more than five turns, or the stars come mostly from channels you pushed yourself, with no organic issues or meaningful forks. That means people don't actually want "watch AI film my story" that badly. Stop. [GitHub repo ↗](https://github.com/Murea-Labs/InWorld) **Research notes: competitor table, acquisition evidence, full research report** Competitors | Name | Where | Launched | Scale | How it differs from this one | |---|---|---|---|---| | Ant LingBot-World | Open source | 2026-01 | 4571 star | Real-time controllable world exploration; no explicit story state or dialogue audio | | Kunlun Matrix-Game | Open source | 2025-05 | 2855 star | Keyboard/mouse-controlled, game-style world model; no text storytelling | | Tencent HunyuanWorld-1.0 / GameCraft | Open source | 2025-07 | 2961 / 767 star | 3D scene or game video generation | | Lightricks LTX-2 | Open source | 2026-01 | 10101 star | The base InWorld runs on; InWorld adds multi-reference images, continuation, and an agent on top | | Alibaba HappyOyster / Shengshu Vidu S2 / Aishi PixVerse R2 | Big-company features | 2026 | Commercial products | Closed source, with real users and distribution channels | | Google Genie 3 / World Labs Marble | Big-company features | 2025-2026 | Industry benchmarks | Explorable worlds, not focused on story agents | | AI Dungeon and other text interactive stories | Web | 2019 | Millions of users | Text only; InWorld turns the story into video | Where users come from - **Verified** **Media feature (Synced)** : 10-18 article "Behind the Dual #1 Rankings…", 10-25 video (92 views on YouTube), 10-27 X post (tagged #PapersAccepted) - **Verified** **Fork and star farming** : 259 of the forks, 229 of them created at :02/:22/:42, about 20 a day starting from 10-25; the accounts have 0 followers, and forks were renamed to random strings; the same account pool is also farming monid-ai/monid and zhiyongmm/diurnix; 0 watchers, 0 HF downloads - **Verified** **Aggregator reposts** : Tencent News, Sohu, AGI Hunt, kaiyuanbang ## Murea-Labs/InWorld research report (2026-11-06) ### One line An "interactive audio-visual world model" open-sourced by Murea Labs (a GitHub org created in 2026-10-12, likely a Chinese team): Murea-Mind (a world-state agent calling Gemini / OpenAI-compatible / DashScope) plus Murea-Fast (an inference service built on LTX-2.5 for multi-reference, synced-audio video generation). Users type actions, and the system films them as a continuous short story, with support for 2–4 player co-op rooms. **Conclusion: this star curve didn't grow organically.** The only real exposure was one set of Synced articles (10-18 article, 10-25 video, 10-27 X post). Starting 10-25, stars and forks grew on a machine-like rhythm: a fork every 20 minutes, landing right on the hour at :02 / :22 / :42, about 20 a day, all from zero-follower alt accounts, and renamed after forking. Stars held steady at 120–150 a day. The Hugging Face weights show 0 downloads, and watchers are 0. This repo is a case study in "botting plus one media push," not organic growth. ### Data - Data card (GitHub repo): created 2026-10-13; at 10-25 it had 29 stars; from 10-26 to 163, then every day +77 to +153,11-06, reaching 1559 stars / 231 forks; last 7 days +920. - GitHub API (2026-11-06 live check): 1693 stars, 251 forks, 0 open issues, subscribers_count (watchers) = 0. Last push was 10-20; no code updates since. - 5 contributors: ying-2026-dev (35 commits), sjsjhsu, bcright1, yichya, plus 15 commits credited to "Codex" (the website and README were written with Codex). - Hugging Face MureaLabs/InWorld: created 10-15, likes 5, downloads 0, downloadsAllTime 0 (HF API, 11-06). In other words, behind 1600 stars, almost no one actually downloaded the weights. Running them requires 4 H20 GPUs, so the barrier was already high. - The stargazers endpoint returned 404, so the star timeline couldn't be retrieved (cause not verified). The fork list was available and is the main evidence. ### Growth tactics (evidence) 1. **Synced media feature (confirmed)** : 2026-10-18 article "Behind the Dual #1 Rankings, They Aren't Trying to Build the Next World Model" (jiqizhixin.com/articles/2026-10-18-10, reposted by Tencent News and Sohu); 10-25 video on WeChat Channels / YouTube / Bilibili / Zhihu, "A New AI Video Trick, Murea-InWorld" (92 views on YouTube); 10-27 English X post from Synced, tagged #PapersAccepted, from Synced's paper-submission column. The article reads throughout like promotion of "dual #1 rankings," most likely a team submission or sponsored piece (whether it was paid is unverified). After these posts went out, stars barely moved (from about 25 to 29 between 10-18 and 10-25). 2. **Fork and star farming (highly suspicious, close to confirmed)** : - Of the 259 forks, 229 were created at :02 / :22 / :42 (within a few seconds of each other), starting 10-25 00:02:09, steadily about 20 a day (10-27 to 11-06, 19–23 a day); most of the rest came from the 11-06 batch, now on a :04 / :24 / :44 rhythm. - Spot-checked fork accounts (dperell, kenghenhot, quizzlocke, amartuplewmune, jasontrans, craiganko) each have 0 followers, 0 following, and 1–5 repos; forks were renamed to strings like rmoerbmv and fqihxeohhcc, a typical farming-service tactic. - The same accounts also forked other repos, such as monid-ai/monid (6133 stars) and zhiyongmm/diurnix (23 stars but 251 forks), suggesting a shared botting pool serving multiple clients. - star grows very steadily (about 125–135 per day), with none of the 'spike then decay' pattern you get from general media exposure. Watchers are 0, issues 0, and HF downloads 0; none of these line up with 1600 organic stars. 3. **Other channels** : searching x-radar's recent 30 days of corpus for Murea / InWorld returned nothing; no discussion found on Reddit or HN; secondary spread only in aggregator sites like AGI Hunt, the open-source chart (kaiyuanbang), and a personal notes repo (ImChong/Robotics_Notebooks). ### Competitors | Name | Form | Launched | Scale | How it differs from InWorld | |---|---|---|---|---| | Ant Group's LingBot-World (Robbyant/lingbot-world) | Open source | 2026-01 | 4571 star | Real-time controllable video world model, leans toward 'exploring scenes'; no explicit plot state and no sound | | Kunlun Matrix-Game | Open source | 2025-05 | 2855 star | Game-style interactive world model, keyboard/mouse control, no dialogue audio | | Tencent HunyuanWorld-1.0 / GameCraft | Open source | 2025-07 / 08 | 2961 / 767 star | 3D scene generation or game video, not text-driven narrative | | Lightricks LTX-2 (the base under Murea-Fast) | Open source | 2026-01 | 10101 star | General audio-video generation; InWorld layers multi-reference images, continuation, and an agent on top of it | | Alibaba HappyOyster, Shengshu Vidu S2, Aishi PixVerse R2 | Big tech / commercial | 2026 | Commercial products | Closed-source interactive video with real users and distribution channels | | Google Genie 3, World Labs Marble | Big-company features / web | 2025–2026 | Industry benchmarks | Explorable 3D / video worlds, not focused on narrative agents | | AI Dungeon, various AI roleplay apps | Web / App | From ¥2019 | Millions of users | Pure text interactive fiction; InWorld is essentially 'filming' it | InWorld's stated differentiation is 'an agent that keeps the world state under control (characters, items, locations locked by reference images) + video continuation with sound + multiplayer co-op', i.e. a fusion of AI Dungeon and video world models. The direction itself is interesting, but the numbers it has produced have nothing to do with that differentiation: real users almost never see it, and HF downloads are 0. ### Why it 'took off' - The stars are mostly bought: forks arrive on a rhythm, all from alt accounts, and star growth is a constant drip; it appears to share a pool of accounts with other clients. - The Machine Heart (机器之心) coverage gave it some real exposure and endorsement ('#1 on WBench, both boards'). The timing lines up with when the inflation started (10-25 video, 10-27 X post), fitting the 'media piece + star-buying to push GitHub trending' playbook. - Product-wise: it needs 4 H20 GPUs to run, the demo is video only with no online playground, so organic usage is hard to get. ### Visibility Word in small circles: only Machine Heart and other Chinese AI outlets covered it; overseas X, Reddit, and HN have basically no discussion; no revenue, and no productized entry point. ### Takeaways - The 'technical report PDF + polished project page + bilingual README + HF weights + top-tier media placements' cold-start package for AI labs comes together fast (the site and README were written by Codex). - 'Explicit world-state agent + reference images to lock characters + video continuation' is a workable approach to long-narrative consistency, applicable to interactive short drama / AI comic products. - The downside: buying stars for GitHub heat gets caught by simple cross-checks like watcher counts, HF downloads, and fork timestamps. ### Uncertain - Why the stargazers endpoint returns 404 (did GitHub flag the repo?) is unverified; whether the stars were all bought can't be checked one by one, so this is inferred from the fork evidence and the steady curve. - Whether the inflation was bought by the team or is third-party activity is unverified; whether the Machine Heart piece was paid placement is unverified. - Murea Labs' entity (company, funding) is unclear; it appears to have some connection to Meituan LongCat's WBench benchmark, also unverified. ### Sources - GitHub API: repos/Murea-Labs/InWorld, /forks, /contributors, users/ (2026-11-06) - Hugging Face API:huggingface.co/api/models/MureaLabs/InWorld - Machine Heart: [www.jiqizhixin.com/articles/2026-10-18-10](https://www.jiqizhixin.com/articles/2026-10-18-10); [x.com/jiqizhixin/status/2115087119660314967](https://x.com/jiqizhixin/status/2115087119660314967); [www.youtube.com/watch?v=2r0-4YGCo1g](https://www.youtube.com/watch?v=2r0-4YGCo1g) - Project page [murea.ai/](https://murea.ai/); technical report [murea.ai/page/paper/InWorld-technical-report.pdf](https://murea.ai/page/paper/InWorld-technical-report.pdf) - Data card /home/ljr/fanout/why/cards/gh_Murea-Labs__InWorld.json This article was written by AI from public data and the data we collect daily. Numbers are sourced, and inferences are labeled.