The best Hacker News stories from Show from the past day
Latest posts:
Show HN: Nobody blocks a keyboard – Rooted Android as a physical USB Deck
Hey HN,
I built Emittr because, i frequently deal with air-gapped or locked-down systems where USB storage is disabled and third party software cannot be installed.
Nobody blocks a physical keyboard, so i used the Linux Kernel's USB gadget subsystem (ConfigFS at /dev/hidg0) to turn a rooted android phone to a real USB HID Keyboard and mouse.
1. No installation on target device. It talks to motherboard USB controller, so it works in every locked down OS as well, BIOS, GRUB, etc.
2. Out of band bridge : we don't store passwords or authenticators on rooted phone. Emittr exposes a web UI on LAN, so we can send keystrokes from our main phones.
3. Types long scripts (chatgpt debuggings), SSH keys or passwords at ~1000 WPM without touching the host's clipboard.
4. OS can see this only as a genuine HID keyboard/mouse. Zero host footprint as no drivers, software or background scripts on target machine.
5. Uses USB-IF 0x0238 via on-screen joysticks instead of synthetic Shift + Scroll hacks.<p>Feel free to check the work and try if you have a rooted android handy.
GitHub: <a href="https://github.com/mr-anjaneyam/Emittr" rel="nofollow">https://github.com/mr-anjaneyam/Emittr</a>
Webpage: <a href="https://emittr.srianjaneyam.me/" rel="nofollow">https://emittr.srianjaneyam.me/</a><p>I'm planning to record and add few demos in various scenarios (Windows, macOS, android, linux, bios) and devices (Phone, PC, TV, STB, Kiosk).<p>Happy to answer any questions and would love to hear your suggestions, feature ideas or edge cases.
Show HN: Mastra Factory, the open-source software factory we use to build Mastra
Hi HN, we're Shane, Sam, and Abhi.<p>Almost two years ago, we shared Mastra, our open-source TypeScript agent framework, on Show HN: <a href="https://news.ycombinator.com/item?id=43103073">https://news.ycombinator.com/item?id=43103073</a>. It got a few hundred upvotes; a couple of weeks later our downloads had doubled, and we were off to the races.<p>Today we're sharing Mastra Factory, the open-source software factory we built to help maintain Mastra itself.<p>At Mastra, we now average nearly 4M downloads per week, and with Factory our team now merges hundreds of PRs a week across the mastra-ai GitHub org. Last week, Factory opened and merged 238 PRs, alongside 630 merged PRs from humans.<p>There's a lot of debate about what a software factory is. For us Factory is a configurable agentic kanban board that's @-taggable in Slack.<p>Under the hood, it uses Mastra’s durable agents for long-running sessions, our conversational memory system, a markdown based wiki with source citations to incorporate company knowledge, signals that subscribe to PR comments wired up into sessions, and a customizable cloud sandbox with our coding harness (working on supporting Claude Code, Codex etc).<p>When we started out, we modeled Factory around our existing workflow-oriented process (triage, plan, build, review, shipped). Three months later, we find ourselves mostly driving it through Slack vs a web UI.<p>We’ve seen more senior engineers kick off sessions in parallel (“triage the 20 oldest issues and let’s evaluate…”). We’re using bug and vuln scanners now, and we’ve closed our repo to outside contributors, asking folks to write detailed issue repros instead.<p>You can get started with `npm create factory`.<p>It’s fully open-source, so you can run it entirely on your infra (we also offer a hosted version). Comment here, we would love to know what you think!<p><a href="https://factory.mastra.ai">https://factory.mastra.ai</a><p><a href="https://github.com/mastra-ai/template-factory" rel="nofollow">https://github.com/mastra-ai/template-factory</a>
Show HN: Mastra Factory, the open-source software factory we use to build Mastra
Hi HN, we're Shane, Sam, and Abhi.<p>Almost two years ago, we shared Mastra, our open-source TypeScript agent framework, on Show HN: <a href="https://news.ycombinator.com/item?id=43103073">https://news.ycombinator.com/item?id=43103073</a>. It got a few hundred upvotes; a couple of weeks later our downloads had doubled, and we were off to the races.<p>Today we're sharing Mastra Factory, the open-source software factory we built to help maintain Mastra itself.<p>At Mastra, we now average nearly 4M downloads per week, and with Factory our team now merges hundreds of PRs a week across the mastra-ai GitHub org. Last week, Factory opened and merged 238 PRs, alongside 630 merged PRs from humans.<p>There's a lot of debate about what a software factory is. For us Factory is a configurable agentic kanban board that's @-taggable in Slack.<p>Under the hood, it uses Mastra’s durable agents for long-running sessions, our conversational memory system, a markdown based wiki with source citations to incorporate company knowledge, signals that subscribe to PR comments wired up into sessions, and a customizable cloud sandbox with our coding harness (working on supporting Claude Code, Codex etc).<p>When we started out, we modeled Factory around our existing workflow-oriented process (triage, plan, build, review, shipped). Three months later, we find ourselves mostly driving it through Slack vs a web UI.<p>We’ve seen more senior engineers kick off sessions in parallel (“triage the 20 oldest issues and let’s evaluate…”). We’re using bug and vuln scanners now, and we’ve closed our repo to outside contributors, asking folks to write detailed issue repros instead.<p>You can get started with `npm create factory`.<p>It’s fully open-source, so you can run it entirely on your infra (we also offer a hosted version). Comment here, we would love to know what you think!<p><a href="https://factory.mastra.ai">https://factory.mastra.ai</a><p><a href="https://github.com/mastra-ai/template-factory" rel="nofollow">https://github.com/mastra-ai/template-factory</a>
Show HN: Mastra Factory, the open-source software factory we use to build Mastra
Hi HN, we're Shane, Sam, and Abhi.<p>Almost two years ago, we shared Mastra, our open-source TypeScript agent framework, on Show HN: <a href="https://news.ycombinator.com/item?id=43103073">https://news.ycombinator.com/item?id=43103073</a>. It got a few hundred upvotes; a couple of weeks later our downloads had doubled, and we were off to the races.<p>Today we're sharing Mastra Factory, the open-source software factory we built to help maintain Mastra itself.<p>At Mastra, we now average nearly 4M downloads per week, and with Factory our team now merges hundreds of PRs a week across the mastra-ai GitHub org. Last week, Factory opened and merged 238 PRs, alongside 630 merged PRs from humans.<p>There's a lot of debate about what a software factory is. For us Factory is a configurable agentic kanban board that's @-taggable in Slack.<p>Under the hood, it uses Mastra’s durable agents for long-running sessions, our conversational memory system, a markdown based wiki with source citations to incorporate company knowledge, signals that subscribe to PR comments wired up into sessions, and a customizable cloud sandbox with our coding harness (working on supporting Claude Code, Codex etc).<p>When we started out, we modeled Factory around our existing workflow-oriented process (triage, plan, build, review, shipped). Three months later, we find ourselves mostly driving it through Slack vs a web UI.<p>We’ve seen more senior engineers kick off sessions in parallel (“triage the 20 oldest issues and let’s evaluate…”). We’re using bug and vuln scanners now, and we’ve closed our repo to outside contributors, asking folks to write detailed issue repros instead.<p>You can get started with `npm create factory`.<p>It’s fully open-source, so you can run it entirely on your infra (we also offer a hosted version). Comment here, we would love to know what you think!<p><a href="https://factory.mastra.ai">https://factory.mastra.ai</a><p><a href="https://github.com/mastra-ai/template-factory" rel="nofollow">https://github.com/mastra-ai/template-factory</a>
Show HN: Take Voyager 1 apart, down to a single memory chip
Voyager 1 reached one light-day from Earth today.<p>On the site you can:<p>- see where both Voyagers are right now<p>- open a nuclear battery to its 24 plutonium spheres<p>- go into the data computer, down to a single memory chip<p>- see how the Golden Record's cover explains how to play it<p>- watch how Voyager 1 was fixed by radio<p>- see which dish on Earth is listening to them, live<p>I also made a 4-minute film from it: <a href="https://www.youtube.com/watch?v=kpzsRo-c1_A" rel="nofollow">https://www.youtube.com/watch?v=kpzsRo-c1_A</a>
Show HN: Take Voyager 1 apart, down to a single memory chip
Voyager 1 reached one light-day from Earth today.<p>On the site you can:<p>- see where both Voyagers are right now<p>- open a nuclear battery to its 24 plutonium spheres<p>- go into the data computer, down to a single memory chip<p>- see how the Golden Record's cover explains how to play it<p>- watch how Voyager 1 was fixed by radio<p>- see which dish on Earth is listening to them, live<p>I also made a 4-minute film from it: <a href="https://www.youtube.com/watch?v=kpzsRo-c1_A" rel="nofollow">https://www.youtube.com/watch?v=kpzsRo-c1_A</a>
Show HN: Take Voyager 1 apart, down to a single memory chip
Voyager 1 reached one light-day from Earth today.<p>On the site you can:<p>- see where both Voyagers are right now<p>- open a nuclear battery to its 24 plutonium spheres<p>- go into the data computer, down to a single memory chip<p>- see how the Golden Record's cover explains how to play it<p>- watch how Voyager 1 was fixed by radio<p>- see which dish on Earth is listening to them, live<p>I also made a 4-minute film from it: <a href="https://www.youtube.com/watch?v=kpzsRo-c1_A" rel="nofollow">https://www.youtube.com/watch?v=kpzsRo-c1_A</a>
Show HN: Type Fighter 2
Type Fighter II is an arcade inspired game for people learning to touch type.<p>Free, no account, and progress stays local. You can try on mobile, best with a physical keyboard.
Show HN: Type Fighter 2
Type Fighter II is an arcade inspired game for people learning to touch type.<p>Free, no account, and progress stays local. You can try on mobile, best with a physical keyboard.
Show HN: Type Fighter 2
Type Fighter II is an arcade inspired game for people learning to touch type.<p>Free, no account, and progress stays local. You can try on mobile, best with a physical keyboard.
Show HN: ExitPath – Open-source ways out of Slack, Notion and Microsoft 365
Show HN: ExitPath – Open-source ways out of Slack, Notion and Microsoft 365
Show HN: ExitPath – Open-source ways out of Slack, Notion and Microsoft 365
Show HN: Birds as Code
Show HN: Agav – an open-source terminal agent designed to waste fewer tokens
Show HN: Open-source, real-time object tracking for live video feeds
I type "forklift" on a live warehouse camera, and tamserve finds and tracks both forklifts, including the one driving in. Then I add "person" without stopping it.<p>tamserve runs Meta's SAM 3 on live camera feeds. SAM 3's own tracker handles one recorded video at a time, runs at about 8 fps, and runs out of memory on long videos. tamserve uses SAM 3 to find objects every few frames and a lightweight tracker (EdgeTAM) to follow them in between, with one copy of each model shared across all your cameras.<p>*On one RTX 4090:*<p>* *Any object you name:* 4x faster than SAM 3's tracker, at 92% of its accuracy
* *People:* 16 cameras at once, 14x less GPU per frame, and slightly more accurate than SAM 3 (89.6 vs 87.8 HOTA)
* *Memory:* flat at 3.5 GB on a 3,000 frame video, where SAM 3 runs out at frame 423<p>*Worth a look if you're building:*<p>* Warehouse or factory safety (forklifts, pallets, hard hats)
* Security or retail camera analytics
* Home or elder care monitoring
* Robotics, wildlife or lab monitoring
* Annotation tools for long videos<p>`pip install "tamserve[sam3]"`<p>Please give me any and all feedback or edit tamserve to meet your needs, and a star helps if you find it useful!
Show HN: Open-source, real-time object tracking for live video feeds
I type "forklift" on a live warehouse camera, and tamserve finds and tracks both forklifts, including the one driving in. Then I add "person" without stopping it.<p>tamserve runs Meta's SAM 3 on live camera feeds. SAM 3's own tracker handles one recorded video at a time, runs at about 8 fps, and runs out of memory on long videos. tamserve uses SAM 3 to find objects every few frames and a lightweight tracker (EdgeTAM) to follow them in between, with one copy of each model shared across all your cameras.<p>*On one RTX 4090:*<p>* *Any object you name:* 4x faster than SAM 3's tracker, at 92% of its accuracy
* *People:* 16 cameras at once, 14x less GPU per frame, and slightly more accurate than SAM 3 (89.6 vs 87.8 HOTA)
* *Memory:* flat at 3.5 GB on a 3,000 frame video, where SAM 3 runs out at frame 423<p>*Worth a look if you're building:*<p>* Warehouse or factory safety (forklifts, pallets, hard hats)
* Security or retail camera analytics
* Home or elder care monitoring
* Robotics, wildlife or lab monitoring
* Annotation tools for long videos<p>`pip install "tamserve[sam3]"`<p>Please give me any and all feedback or edit tamserve to meet your needs, and a star helps if you find it useful!
Show HN: VennDB: append-only in-memory Rust DB for rows queried using bit flags
Show HN: TrnRT – An open-source LLM inference engine for AWS Trainium
Hi HN, Sanchit here from RunAnywhere (YC W26). We just open-sourced TrnRT: Trainium RunTime, our inference engine for AWS Trainium. We started 3 weeks back, and currently we’re able to serve and run GLM 5.3 flash (FP8) on AWS trn2.48xlarge, and will bring more models support soon. It’s open to contributions :)<p>What it does today:<p>- Serves 2 instances of GLM-5.3 flash on a single box, 8 trn2 with a cache aware router.<p>- Can process 3.57M TPM (tokens per minute) on SemiAnalysis's AgentX benchmarks.<p>- Runs decode as one NKI kernel per layer half, or for the 11 sparse-attention layers fuses both blocks into a single kernel, for faster performance.<p>We started working on Trn2 because we wanted to explore it and see if we can come close to beating it’s performance in terms of cost efficiency compared to H100/H200. A big MoE with 1M context window such as GLM-5.3 flash requires a lot of memory, and memory/tb on a trn2.48xlarge costs less than half compared to 8x h200 on AWS, and hence we wanted to experiment with it and now we’re sharing some early results.<p>But AWS's own stack didn't support the model: vLLM Neuron's latest release did not register any GLM model, and NxDI, AWS's own distributed inference framework, is no longer under active development. We also saw the drop of SGLang Neuron implementation yesterday, but there is no good support for Trn2, which is the only chip available for public use right now.<p>We ran semiAnalysis's AgentX benchmarks, which replays real coding agent sessions:<p>- 1 box served 28.9 RPM(requests per minute) with 32 agents at batch 12 (also known as concurrency)<p>-The P50 TTFT with a cached prompt at 16 agents is ~1.24 seconds with a decode of 35.9 tps.<p>Initially, the recurrence in one kda (linear attention) layer was ~2% of the layer's attention time at batch 1, and the rest was projections, transposes and XLA glue between small kernels.<p>So we wrote the megakernel so that it covers the whole layer in 2 kernel dispatch, which made decode 25% faster at batch 8.<p>The sparse attention kernel keep a layer’s working weights in SBUF, distributing weighted streaming across the cores and handling the interchip communication within<p>There’s an approximation in the sparse-attention prefill, affecting about 1.2% of block selections.<p>You can refer to the write-up for the profiling results:<p><a href="https://runanywhere.ai/blog/glm-5-3-flash-trainium2-nki-megakernels">https://runanywhere.ai/blog/glm-5-3-flash-trainium2-nki-mega...</a><p>Please give it a try yourself on a trn2.48xlarge and reproduce the AgentX results by following the steps in docs/RESULTS.md. Happy to collaborate and hear any feedback.<p>Deploying high performant dedicated inference at lab like performance and half the price for open models such as the GLM-5.3 family, Kimi K3, Deepseek flash etc. is how we make money. You can talk to us regarding the same at founders@runanywhere.ai.<p>We hope you find this useful. If you are working with Trainium or have experience with NKI kernels, we'd love to get your honest feedback. Feel free to share the models you want support for next, and we welcome any contributions to the repository.
Show HN: TrnRT – An open-source LLM inference engine for AWS Trainium
Hi HN, Sanchit here from RunAnywhere (YC W26). We just open-sourced TrnRT: Trainium RunTime, our inference engine for AWS Trainium. We started 3 weeks back, and currently we’re able to serve and run GLM 5.3 flash (FP8) on AWS trn2.48xlarge, and will bring more models support soon. It’s open to contributions :)<p>What it does today:<p>- Serves 2 instances of GLM-5.3 flash on a single box, 8 trn2 with a cache aware router.<p>- Can process 3.57M TPM (tokens per minute) on SemiAnalysis's AgentX benchmarks.<p>- Runs decode as one NKI kernel per layer half, or for the 11 sparse-attention layers fuses both blocks into a single kernel, for faster performance.<p>We started working on Trn2 because we wanted to explore it and see if we can come close to beating it’s performance in terms of cost efficiency compared to H100/H200. A big MoE with 1M context window such as GLM-5.3 flash requires a lot of memory, and memory/tb on a trn2.48xlarge costs less than half compared to 8x h200 on AWS, and hence we wanted to experiment with it and now we’re sharing some early results.<p>But AWS's own stack didn't support the model: vLLM Neuron's latest release did not register any GLM model, and NxDI, AWS's own distributed inference framework, is no longer under active development. We also saw the drop of SGLang Neuron implementation yesterday, but there is no good support for Trn2, which is the only chip available for public use right now.<p>We ran semiAnalysis's AgentX benchmarks, which replays real coding agent sessions:<p>- 1 box served 28.9 RPM(requests per minute) with 32 agents at batch 12 (also known as concurrency)<p>-The P50 TTFT with a cached prompt at 16 agents is ~1.24 seconds with a decode of 35.9 tps.<p>Initially, the recurrence in one kda (linear attention) layer was ~2% of the layer's attention time at batch 1, and the rest was projections, transposes and XLA glue between small kernels.<p>So we wrote the megakernel so that it covers the whole layer in 2 kernel dispatch, which made decode 25% faster at batch 8.<p>The sparse attention kernel keep a layer’s working weights in SBUF, distributing weighted streaming across the cores and handling the interchip communication within<p>There’s an approximation in the sparse-attention prefill, affecting about 1.2% of block selections.<p>You can refer to the write-up for the profiling results:<p><a href="https://runanywhere.ai/blog/glm-5-3-flash-trainium2-nki-megakernels">https://runanywhere.ai/blog/glm-5-3-flash-trainium2-nki-mega...</a><p>Please give it a try yourself on a trn2.48xlarge and reproduce the AgentX results by following the steps in docs/RESULTS.md. Happy to collaborate and hear any feedback.<p>Deploying high performant dedicated inference at lab like performance and half the price for open models such as the GLM-5.3 family, Kimi K3, Deepseek flash etc. is how we make money. You can talk to us regarding the same at founders@runanywhere.ai.<p>We hope you find this useful. If you are working with Trainium or have experience with NKI kernels, we'd love to get your honest feedback. Feel free to share the models you want support for next, and we welcome any contributions to the repository.