AI Repo: comet-ml/opik

Skyler: Hey everyone, welcome back to Git News A I, the show where we dig into the hottest A I repositories blowing up on GitHub right now. Quick shoutout to our sponsor LimitLess A I — more on them at the end of the episode. Today we've got five absolutely wild repos to cover, from L L M debugging tools to A I-powered video generation, persistent agent memory, auto-generated PowerPoints, and an open source coding agent that's racking up stars like crazy. Let's get into it. Skyler: Alright, kicking things off with a repo called Opik from Comet M L, sitting at just under twenty thousand stars. If you've ever shipped an L L M application and thought "I have absolutely no idea why this thing is behaving the way it is," this tool is basically made for you. Opik is a full-featured platform for debugging, evaluating, and monitoring your L L M applications, your R A G systems, and agentic workflows. So think of it as your observability layer specifically designed for the messy world of A I apps. Skyler: What makes it really compelling is the combination of comprehensive tracing — so you can actually follow what's happening at every step of your pipeline — automated evaluations, and production-ready dashboards. That last part is huge. A lot of observability tools feel like they're built for demos, not for real production environments. Opik is clearly trying to close that gap. Skyler: The tracing capabilities mean you can capture inputs, outputs, and metadata across every chain, every agent call, every retrieval step. You can then run automated evaluations against those traces to check things like hallucination rates, answer relevance, context precision — all the metrics that actually matter when you're trying to ship reliable A I products. Skyler: If you're building anything with L L Ms — whether that's a chatbot, a document Q and A system, or a multi-agent workflow — and you want real visibility into what's happening under the hood, Opik is absolutely worth a look. Check the show notes for the link. Skyler: Next up, and this one has a massive star count — we're talking nearly ninety-four thousand stars — it's MoneyPrinterTurbo. The name alone should tell you something. This is a Python project that lets you generate high-quality short videos with basically one click using A I L L Ms. And yeah, it's as cool as it sounds. Skyler: The idea is simple but powerful. You feed it a topic or a script, and it handles the rest — sourcing footage, generating narration, adding subtitles, and stitching everything together into a ready-to-post short video. We're talking the kind of content you'd see on TikTok, YouTube Shorts, or Instagram Reels. Skyler: There was some recent activity here too — a pull request got merged that adds TwelveLabs integration, which is super interesting. TwelveLabs is a multimodal video understanding A I platform, so this integration likely opens up smarter video search and content matching capabilities inside the pipeline. Skyler: What I love about this project is that it democratizes video content creation in a real way. You don't need a video editor, you don't need a voiceover artist, you don't need a stock footage subscription. The A I handles the heavy lifting. Marketers, content creators, developers building content automation tools — this one's a goldmine. Check the show notes for the link. Skyler: Okay, repo number three is cognee, and this one is doing something really fundamental that the A I agent space desperately needs. Cognee is an open-source A I memory platform for agents, sitting at over twenty-four thousand stars. The core problem it's solving is that most A I agents are basically amnesiac — every session starts from scratch, and there's no persistent understanding of who the user is, what's happened before, or what knowledge the agent should be drawing on. Skyler: Cognee fixes this by giving your A I agents persistent long-term memory across sessions using a self-hosted knowledge graph engine. So instead of stuffing everything into a context window and hoping for the best, you're building a structured, queryable graph of knowledge that the agent can actually reason over. Skyler: The self-hosted aspect is a big deal for teams that care about data privacy and control. You're not shipping your user data off to some third-party memory service — you own the infrastructure. And being open-source means you can inspect, customize, and extend it however you need. Skyler: Think about the use cases here — personal A I assistants that actually remember your preferences, customer support agents that build context over time, coding assistants that understand the history of your project. This is the kind of foundational infrastructure that makes agentic A I actually useful in the long run. If you're building agents and you haven't thought about memory yet, cognee is the perfect place to start. Check the show notes for the link. Skyler: Alright, repo four is ppt-master by Hugo He, and this one just crossed thirty-three thousand stars which is honestly impressive for how niche it sounds. The pitch is: give it any document, and A I generates a real, fully editable PowerPoint file. Not a PDF of slides. Not slide images. An actual native P P T X file with shapes, animations, speaker notes, and even audio narration. Skyler: Let me break down why that distinction matters. A lot of A I slide tools generate images of slides, which means once it's generated, you can't really edit it in PowerPoint without starting over. ppt-master generates actual native PowerPoint elements — so you get real shapes you can move around, real text boxes you can edit, real animations you can tweak. It's treating P P T X as a first-class output format. Skyler: The speaker notes voiced as audio narration is a genuinely clever touch. So you could potentially generate a full presentation with a voiceover baked in without recording anything yourself. And the ability to use your own P P T X template means you can enforce brand guidelines — your fonts, your color palette, your slide layouts. Skyler: There's been some recent activity adding brand presets for specific companies, which suggests this is being used in real enterprise contexts. If you're in a role where you produce a lot of presentations — analysts, consultants, developers writing technical docs — or if you're building document-to-presentation tooling, this repo is seriously worth exploring. Check the show notes for the link. Skyler: And saving the biggest for last — opencode. One hundred and seventy-nine thousand stars. That is not a typo. This is an open-source coding agent written in TypeScript, and the community has absolutely rallied around it. The description is refreshingly simple: "the open source coding agent." No frills, just vibes. Skyler: What we know is that it's a coding agent, meaning it's designed to autonomously understand, write, and modify code — not just autocomplete, but actually reason about what needs to happen and take action. The TypeScript stack is interesting because it means this is potentially more accessible to web developers and full-stack engineers who might not be as comfortable with Python-heavy M L tooling. Skyler: The recent commits show active development — bug fixes and project updates are coming in regularly, which is a great sign for a project this young in terms of its GitHub trajectory. Nearly a hundred and eighty thousand stars is the kind of number you see when a project really hits a nerve with the developer community. Skyler: If you've been following the coding agent space — things like Devin, or the various open-source alternatives that have been popping up — opencode is clearly one of the most-watched projects right now. Whether you want to use it, contribute to it, or just study how a coding agent gets built, this is a repo you need to have on your radar. Check the show notes for the link. Skyler: And that is a wrap on today's episode of Git News A I. Huge thanks to LimitLess A I for sponsoring the show — seriously, go check them out if you want to level up your A I workflows. All the repos we talked about today are linked in the show notes so you can dive in right after this. Skyler: If you found this episode useful, please subscribe wherever you get your podcasts, leave us a review if you're feeling generous, and share it with a developer friend who would appreciate the deep dive. We'll be back soon with more of the freshest A I repositories on GitHub. I'm Skyler, and I'll catch you next time.