Show HN: Plandex v2 – open source AI coding agent for large projects and tasks https://ift.tt/8b6arfS

Show HN: Plandex v2 – open source AI coding agent for large projects and tasks Hey HN! I’m Dane, the creator of Plandex ( https://ift.tt/DX97RW3 ), an open source AI coding agent focused especially on tackling large tasks in real world software projects. You can watch a 2 minute demo of Plandex in action here: https://www.youtube.com/watch?v=SFSu2vNmlLk And here’s more of a tutorial style demo showing how Plandex can automatically debug a browser application: https://www.youtube.com/watch?v=VCegxOCAPq0 I launched Plandex v1 here on HN a little less than a year ago ( https://ift.tt/mlYbwcD ). Now I’m launching a major update, Plandex v2, which is the result of 8 months of heads down work, and is in effect a whole new project/product. In short, Plandex is now a top-tier coding agent with fully autonomous capabilities. It combines models from Anthropic, OpenAI, and Google to achieve better results, more reliable agent behavior, better cost efficiency, and better performance than is possible by using only a single provider’s models. I believe it is now one of the best tools available for working on large tasks in real world codebases with AI. It has an effective context window of 2M tokens, and can index projects of 20M tokens and beyond using tree-sitter project maps (30+ languages are supported). It can effectively find relevant context in massive million-line projects like SQLite, Redis, and Git. A bit more on some of Plandex’s key features: - Plandex has a built-in diff review sandbox that helps you get the benefits of AI without leaving behind a mess in your project. By default, all changes accumulate in the sandbox until you approve them. The sandbox is version-controlled. You can rewind it to any previous point, and you can also create branches to try out alternative approaches. - It offers a ‘full auto mode’ that can complete large tasks autonomously end-to-end, including high level planning, context loading, detailed planning, implementation, command execution (for dependencies, builds, tests, etc.), and debugging. - The autonomy level is highly configurable. You can move up and down the ladder of autonomy depending on the task, your comfort level, and how you weigh cost optimization vs. effort and results. - Models and model settings are also very configurable. There are built-in models and model packs for different use cases. You can also add custom models and model packs, and customize model settings like temperature or top-p. All model changes are version controlled, so you can use branches to try out the same task with different models. The newly released OpenAI models and the paid Gemini 2.5 Pro model will be integrated in the default model pack soon. - It can be easily self-hosted, including a ‘local mode’ for a very fast local single-user setup with Docker. - Cloud hosting is also available for added convenience with a couple of subscription tiers: an ‘Integrated Models’ mode that requires no other accounts or API keys and allows you to manage billing/budgeting/spending alerts and track usage centrally, and a ‘BYO API Key’ mode that allows you to use your own OpenAI/OpenRouter accounts. I’d love to get more HNers in the Plandex Discord ( https://ift.tt/lVIwM4K ). Please join and say hi! And of course I’d love to hear your feedback, whether positive or negative. Thanks so much! https://ift.tt/DX97RW3 April 17, 2025 at 02:56AM

Show HN: Rocal – Full-stack WASM framework https://ift.tt/F3wEDU9

Show HN: Rocal – Full-stack WASM framework https://ift.tt/x4GohYI April 16, 2025 at 06:43AM

Show HN: Particle - News, Organized https://ift.tt/zGZfXMS

Show HN: Particle - News, Organized Hello HN! Particle News product engineer here. Keeping up with the news is overwhelming in an age of information overload. Particle reimagines the experience by organizing articles into comprehensive "Stories," offering clear, concise summaries to quickly grasp what matters. Today, we reached the #1 spot in "Newspapers & Magazine" on the iOS App Store—and I thought I'd share a bit of our backstory. I've been connected to this team for a long time. About 20 years ago, I shared a house with our CTO and co-founder Marcel Molina. I helped him get started with programming. Since then, Marcel has had an extraordinary career—becoming a senior staff engineer at Twitter, where he helped build foundational features like Retweets, Notifications, and Lists, and later working at Tesla on manufacturing execution systems that scaled across Gigafactories. At Twitter, Marcel worked closely with our CEO and product visionary Sara Beykpour, who led initiatives like Twitter Blue, Twitter Video, and the experimental app twttr . Sara has a background in Software Engineering and Cognitive Science from the University of Waterloo and spent over a decade at Twitter in engineering and leadership roles. In late 2022, Sara and Marcel started prototyping a news app that could reduce the cognitive and emotional burden of staying informed—by using AI to help people understand more, faster. They were soon joined by a few other former Twitter colleagues who helped shape the early concept into a working iOS application. I joined about 15 months ago to contribute across the entire stack. Since then, I've helped design and build major iOS features, rewritten our public website on Cloudflare Workers, and implemented new functionality in our Go backend, which is driven by Google Cloud's Pub/Sub architecture. What Makes Particle News Different Particle helps you navigate the news effortlessly—leveraging AI to help you understand more, faster. Some highlights: • Personalized News – Your feed is tailored to your interests. You can follow specific people, places, and things so you never miss what matters to you. • Clear Summaries – Get a quick overview or dive deeper with detailed, structured context—summarized in natural language. • Perspective Tools – Features like "Opposite Sides" and our political spectrum chart let you explore stories through multiple lenses. • Interactive Q&A – Ask questions about any story and get concise answers with sources and citations. • Audio Summaries – Use the "Play" feature to listen to your feed, specific stories, or even select articles—great for hands-free or on-the-go moments. One of the things we're most proud of is how Particle supports publishers. We've partnered with outlets like Reuters, AFP, and Fortune to host some of their content via APIs. These partners get prominent placement, and their links are highlighted in gold to stand out. This model aims to drive traffic back to publishers and reward high-quality journalism, rather than just aggregating and commodifying it. Transparency is a core value: all sources are cited, generated answers are grounded in evidence, and we take real care to prevent AI hallucinations or misleading summaries. Despite negligible marketing spend, Particle has grown to the top of its category by focusing on engagement with early users and meaningful partnerships with the media ecosystem. Coming soon: weekday mini crosswords—a new feature designed by another longtime friend of ours from 20 years back who went on to work at Twitter, lead development on Firewatch, and release his own games independently. It's incredibly fun and rewarding to be building something meaningful with old and new friends. I feel lucky every day to work alongside some of the best product, design, and engineering minds on a project we hope will help people stay engaged with democracy without burning out. https://ift.tt/rqEpCWw April 16, 2025 at 02:56AM

Show HN: Torque – A lightweight meta-assembler for any processor https://ift.tt/bINhzVe

Show HN: Torque – A lightweight meta-assembler for any processor Hello everyone, I've been working on this project for the past few months. Torque is a meta-assembler: instead of having an instruction set built into the assembler, you use macros to build up a small language that decribes an instruction set and then you use that to write your program. It's designed to work for any microcontroller/processor architecture, you build from the bit level upwards so there aren't any assumptions around word widths, instruction formats, or endianness. I created Torque initially to write programs for a PIC microcontroller, after running into difficulties with the official assembler. I've also used it to write programs for the Z80 processor inside an old TRS-80 computer. Let me know if you try it out or have any questions! https://ift.tt/v5xBrUn April 16, 2025 at 03:16AM

Show HN: I asked Gemini2.5 pro to create a "bro version" of a stoic book https://ift.tt/rBkaOjT

Show HN: I asked Gemini2.5 pro to create a "bro version" of a stoic book https://ift.tt/0ZUnv3k April 15, 2025 at 11:06PM

Show HN: ElephantShadow: PHP for SSR of Webcomponents in declarative shadow DOM https://ift.tt/38plwJs

Show HN: ElephantShadow: PHP for SSR of Webcomponents in declarative shadow DOM https://ift.tt/xVfKcha April 14, 2025 at 11:23PM

Show HN: ActorCore – Stateful serverless framework that runs anywhere https://ift.tt/bBDifG1

Show HN: ActorCore – Stateful serverless framework that runs anywhere TL;DR: ActorCore is a stateful serverless framework that can be deployed to Rivet, Cloudflare, Bun, Node.js and more. It's the easiest way to build stateful, AI agent, collaborative, or local-first applications. Hey HN! A few months ago we launched Rivet Actors ( https://ift.tt/xRIHzXo ) as an open-source alternative to Cloudflare Durable Objects. Shortly after launching, we realized our goal is not to win over existing Durable Objects developers, but rather to grow the stateful serverless ecosystem. For context – "stateful serverless" is effectively the actor model with persistent state attached. Think Lambda functions with local storage & runs indefinitely. It's a a technology that’s gaining traction to ship faster, achieve higher performance, and outscale Postgres. The most widely used implementation is Cloudflare Durable Objects. In the process of talking to developers, we kept hearing three common concerns: - Vendor lock-in: Developers are hesitant to adopt a new programming model if there's no clear off-ramp. While it's straightforward to migrate a Postgres database, stateful serverless platforms like Rivet Actors or Durable Objects can feel locked-in due to lack of viable alternatives. - Ecosystem: Choosing a well-known database like Postgres comes with a mature ecosystem. Adopting a new model means rebuilding tooling and patterns from scratch. - Conceptual gap: Many developers have spent their entire careers designing systems with intentionally separated state and compute. A model that merges the two can feel backwards at first. After hearing these concerns again and again, we came to the conclusion that the best solution was to build a framework that works with as many platforms as possible to reduce lock-in (concern #1) and grow a shared ecosystem of tools (concern #2). It turns out, we already had a battle-tested framework built on top of Rivet Actors that we’ve been using for years. (It has a long, funky history beyond Rivet in gaming I won't get into here.) Thus, we split out the framework in to a new repo, added support for four platforms (easier said than done), and called it ActorCore. It gives developers multiple platforms to choose from when adopting stateful serverless and creates a foundation for a broader, cross-platform ecosystem. However, this still leaves concern #3: the conceptual gap. While this isn't something we can solve with a framework, I personally spend ~40% of my time working on docs, content, and examples to help resolve this. ActorCore is also panning out to be community-driven as hoped, which enables more people to try and share their experience with stateful serverless. Give ActorCore a try, read the roadmap, and let us know where we can improve documentation. If you're hesitant about trying stateful serverless, I'd love to learn more in the comments. Consider giving us a star on GitHub: https://ift.tt/CnFk7ND https://ift.tt/FKhZEOr April 14, 2025 at 08:27PM