The best Hacker News stories from Show from the past day
Latest posts:
Show HN: My 70 year old grandma is learning to code and made a word game
Show HN: Test your WireGuard connectivity and see global stats, no client needed
Hi HN,<p>Some misbehaving networks drop WireGuard packets either by accident or on purpose. Commonly the latter is done with simple DPI rules that block the handshake initiation [1], but it could be applied to other message types as well.<p>We thought it would be great if there was tool for folks to use as a quick litmus test to see if this happening for them, without having to configure a client to send data through a random, functional WireGuard tunnel to an untrusted remote host. So we built probe.sh.<p>How it works:<p>- The probe.sh web app is an Elixir Phoenix app that spawns a few gen_udp servers across a variety of common UDP ports.
- When a user visits the app, Probe starts a LiveView process and generates a unique cryptographic token to use for the test.
- When the user runs the script shown, it first sends an HTTP request to start the test, followed by a series of UDP payloads, and finally either a complete or cancel request to end the test.
- The UDP payloads are crafted to resemble real world WireGuard packets and sent with widely available tools like netcat (Unix) and System.Net.Sockets.UdpClient (Win) already on your OS.
- The gen_udp server receives these payloads, and if they match one of the four WireGuard message types by header, it broadcasts test updates to the LiveView process for that test, and the test is marked as success.
- The user is immediately shown the results of the test.<p>The entire tool is open source at <a href="https://github.com/firezone/probe">https://github.com/firezone/probe</a> (README contains guide for self-hosting) and you can find a FAQ with more useful info at <a href="https://probe.sh/faq" rel="nofollow">https://probe.sh/faq</a>. You can also see our tally of global results organized by country: <a href="https://probe.sh/stats" rel="nofollow">https://probe.sh/stats</a><p>We hope you find it useful for testing your network for WireGuard connectivity issues.<p>Thanks for reading - feedback welcome!<p>[1] <a href="https://x.com/6h4n3m/status/1459462360003919875" rel="nofollow">https://x.com/6h4n3m/status/1459462360003919875</a>
Show HN: Pg_replicate – Build Postgres replication applications in Rust
Show HN: Pg_replicate – Build Postgres replication applications in Rust
Show HN: Pg_replicate – Build Postgres replication applications in Rust
Show HN: Spawn – Build iOS Apps with English
Hi HN,<p>Spawn lets you build apps with human language. You can include images, audio and other files in your app just by dragging and dropping.<p>Here's a demo video where I build the game of Snake in 60 seconds: <a href="https://www.youtube.com/watch?v=qIqp7cvmE_M" rel="nofollow">https://www.youtube.com/watch?v=qIqp7cvmE_M</a><p>As it’s still in alpha, you may need to regenerate a few times to get the app functioning the way you want. So far I’ve found it’s pretty good at generating simple utilities and games. Over time, as both Spawn and the underlying models improve, it will be able to build more complex software.<p>I have a bunch of ideas for future updates, for example:<p>- Allow users to include wireframes that Spawn will follow as it generates their app.<p>- Android support in Mac, Windows & Linux Spawn desktop app. This way, people can use Spawn to build for both mobile platforms using the same app specifications. Imagine someone living in rural India using their Chromebook to build apps for their Android phone, without needing to learn how to code.<p>- Enable people to build apps that have accounts and backend logic by autogenerating and deploying a backend during app generation.<p>Any feedback would be greatly appreciated.
Show HN: Personal Interactive Cantonese Dictionary
Built with Nuxt & MongoDB – a way to preserve language with personal audio, anecdotes and stories. Hope to make a version open source one day so others can preserve their native languages.
Show HN: Personal Interactive Cantonese Dictionary
Built with Nuxt & MongoDB – a way to preserve language with personal audio, anecdotes and stories. Hope to make a version open source one day so others can preserve their native languages.
Show HN: Personal Interactive Cantonese Dictionary
Built with Nuxt & MongoDB – a way to preserve language with personal audio, anecdotes and stories. Hope to make a version open source one day so others can preserve their native languages.
Show HN: I built Mailhub – A scalable API for sending emails with ease not tears
Show HN: I built Mailhub – A scalable API for sending emails with ease not tears
Show HN: I built Mailhub – A scalable API for sending emails with ease not tears
Show HN: Nous – Open-Source Agent Framework with Autonomous, SWE Agents, WebUI
Hello HN!
The day has finally come to stop adding features and start sharing what I've been building the last 5-6 months.<p>It's a bit of CrewAI, OpenDevon, LangFuse/Cloud all in one, providing devs who prefer TypeScript an integrated framework thats provides a lot out of the box to start experimenting and building agents with.<p>It started after peeking at the LangChain docs a few times and never liking the example code. I began experimenting with automating a simple Jira request from the engineering team to add an index to one of our Google Spanner databases (for context I'm the DevOps/SRE lead for an AdTech company).<p>It incudes the tooling we're building out to automate processes from a DevOps/SRE perspective, which initially includes a configurable GitLab merge request AI reviewer.<p>The initial layer above Aider (<a href="https://aider.chat/" rel="nofollow">https://aider.chat/</a>) grew into coding agent and an autonomous agent with LLM-independent function calling with auto-generated function schemas.<p>And as testing via the CLI became unwieldy soon grew database persistence, tracing, a Web UI and human-in-the-loop functionality.<p>One of the more interesting additions is the new autonomous agent which generates Python code that can call the available functions.
Using the pyodide library the tool objects are proxied into the Python scope and executed in a WebAssembly sandbox.<p>As its able to perform multiple calls and validation logic in a single control loop, it can reduce the cost and latency, getting the most out of the frontier LLMs calls with better reasoning.<p>Benchmark runners for the autonomous agent and coding benchmarks are in the works to get some numbers on the capabilities so far. I'm looking forward to getting back to implementing all the ideas around improving the code and autonomous agents from a metacognitive perspective after spending time on docs, refactorings and tidying up recently.<p>Check it out at <a href="https://github.com/trafficguard/nous">https://github.com/trafficguard/nous</a>
Show HN: Nous – Open-Source Agent Framework with Autonomous, SWE Agents, WebUI
Hello HN!
The day has finally come to stop adding features and start sharing what I've been building the last 5-6 months.<p>It's a bit of CrewAI, OpenDevon, LangFuse/Cloud all in one, providing devs who prefer TypeScript an integrated framework thats provides a lot out of the box to start experimenting and building agents with.<p>It started after peeking at the LangChain docs a few times and never liking the example code. I began experimenting with automating a simple Jira request from the engineering team to add an index to one of our Google Spanner databases (for context I'm the DevOps/SRE lead for an AdTech company).<p>It incudes the tooling we're building out to automate processes from a DevOps/SRE perspective, which initially includes a configurable GitLab merge request AI reviewer.<p>The initial layer above Aider (<a href="https://aider.chat/" rel="nofollow">https://aider.chat/</a>) grew into coding agent and an autonomous agent with LLM-independent function calling with auto-generated function schemas.<p>And as testing via the CLI became unwieldy soon grew database persistence, tracing, a Web UI and human-in-the-loop functionality.<p>One of the more interesting additions is the new autonomous agent which generates Python code that can call the available functions.
Using the pyodide library the tool objects are proxied into the Python scope and executed in a WebAssembly sandbox.<p>As its able to perform multiple calls and validation logic in a single control loop, it can reduce the cost and latency, getting the most out of the frontier LLMs calls with better reasoning.<p>Benchmark runners for the autonomous agent and coding benchmarks are in the works to get some numbers on the capabilities so far. I'm looking forward to getting back to implementing all the ideas around improving the code and autonomous agents from a metacognitive perspective after spending time on docs, refactorings and tidying up recently.<p>Check it out at <a href="https://github.com/trafficguard/nous">https://github.com/trafficguard/nous</a>
Show HN: Nous – Open-Source Agent Framework with Autonomous, SWE Agents, WebUI
Hello HN!
The day has finally come to stop adding features and start sharing what I've been building the last 5-6 months.<p>It's a bit of CrewAI, OpenDevon, LangFuse/Cloud all in one, providing devs who prefer TypeScript an integrated framework thats provides a lot out of the box to start experimenting and building agents with.<p>It started after peeking at the LangChain docs a few times and never liking the example code. I began experimenting with automating a simple Jira request from the engineering team to add an index to one of our Google Spanner databases (for context I'm the DevOps/SRE lead for an AdTech company).<p>It incudes the tooling we're building out to automate processes from a DevOps/SRE perspective, which initially includes a configurable GitLab merge request AI reviewer.<p>The initial layer above Aider (<a href="https://aider.chat/" rel="nofollow">https://aider.chat/</a>) grew into coding agent and an autonomous agent with LLM-independent function calling with auto-generated function schemas.<p>And as testing via the CLI became unwieldy soon grew database persistence, tracing, a Web UI and human-in-the-loop functionality.<p>One of the more interesting additions is the new autonomous agent which generates Python code that can call the available functions.
Using the pyodide library the tool objects are proxied into the Python scope and executed in a WebAssembly sandbox.<p>As its able to perform multiple calls and validation logic in a single control loop, it can reduce the cost and latency, getting the most out of the frontier LLMs calls with better reasoning.<p>Benchmark runners for the autonomous agent and coding benchmarks are in the works to get some numbers on the capabilities so far. I'm looking forward to getting back to implementing all the ideas around improving the code and autonomous agents from a metacognitive perspective after spending time on docs, refactorings and tidying up recently.<p>Check it out at <a href="https://github.com/trafficguard/nous">https://github.com/trafficguard/nous</a>
Show HN: Attaching to a virtual GPU over TCP
We developed a tool to trick your computer into thinking it’s attached to a GPU which actually sits across a network. This allows you to switch the number or type of GPUs you’re using with a single command.
Show HN: Attaching to a virtual GPU over TCP
We developed a tool to trick your computer into thinking it’s attached to a GPU which actually sits across a network. This allows you to switch the number or type of GPUs you’re using with a single command.
Show HN: Attaching to a virtual GPU over TCP
We developed a tool to trick your computer into thinking it’s attached to a GPU which actually sits across a network. This allows you to switch the number or type of GPUs you’re using with a single command.
Show HN: LLM-aided OCR – Correcting Tesseract OCR errors with LLMs
Almost exactly 1 year ago, I submitted something to HN about using Llama2 (which had just come out) to improve the output of Tesseract OCR by correcting obvious OCR errors [0]. That was exciting at the time because OpenAI's API calls were still quite expensive for GPT4, and the cost of running it on a book-length PDF would just be prohibitive. In contrast, you could run Llama2 locally on a machine with just a CPU, and it would be extremely slow, but "free" if you had a spare machine lying around.<p>Well, it's amazing how things have changed since then. Not only have models gotten a lot better, but the latest "low tier" offerings from OpenAI (GPT4o-mini) and Anthropic (Claude3-Haiku) are incredibly cheap and incredibly fast. So cheap and fast, in fact, that you can now break the document up into little chunks and submit them to the API concurrently (where each chunk can go through a multi-stage process, in which the output of the first stage is passed into another prompt for the next stage) and assemble it all in a shockingly short amount of time, and for basically a rounding error in terms of cost.<p>My original project had all sorts of complex stuff for detecting hallucinations and incorrect, spurious additions to the text (like "Here is the corrected text" preambles). But the newer models are already good enough to eliminate most of that stuff. And you can get very impressive results with the multi-stage approach. In this case, the first pass asks it to correct OCR errors and to remove line breaks in the middle of a word and things like that. The next stage takes that as the input and asks the model to do things like reformat the text using markdown, to suppress page numbers and repeated page headers, etc. Anyway, I think the samples (which take less than 1-2 minutes to generate) show the power of the approach:<p>Original PDF:
<a href="https://github.com/Dicklesworthstone/llm_aided_ocr/blob/main/160301289-Warren-Buffett-Katharine-Graham-Letter.pdf">https://github.com/Dicklesworthstone/llm_aided_ocr/blob/main...</a><p>Raw OCR Output:
<a href="https://github.com/Dicklesworthstone/llm_aided_ocr/blob/main/160301289-Warren-Buffett-Katharine-Graham-Letter__raw_ocr_output.txt">https://github.com/Dicklesworthstone/llm_aided_ocr/blob/main...</a><p>LLM-Corrected Markdown Output:
<a href="https://github.com/Dicklesworthstone/llm_aided_ocr/blob/main/160301289-Warren-Buffett-Katharine-Graham-Letter_llm_corrected.md">https://github.com/Dicklesworthstone/llm_aided_ocr/blob/main...</a><p>One interesting thing I found was that almost all my attempts to fix/improve things using "classical" methods like regex and other rule based things made everything worse and more brittle, and the real improvements came from adjusting the prompts to make things clearer for the model, and not asking the model to do too much in a single pass (like fixing OCR mistakes AND converting to markdown format).<p>Anyway, this project is very handy if you have some old scanned books you want to read from Archive.org or Google Books on a Kindle or other ereader device and want things to be re-flowable and clear. It's still not perfect, but I bet within the next year the models will improve even more that it will get closer to 100%. Hope you like it!<p>[0] <a href="https://news.ycombinator.com/item?id=36976333">https://news.ycombinator.com/item?id=36976333</a>
Show HN: LLM-aided OCR – Correcting Tesseract OCR errors with LLMs
Almost exactly 1 year ago, I submitted something to HN about using Llama2 (which had just come out) to improve the output of Tesseract OCR by correcting obvious OCR errors [0]. That was exciting at the time because OpenAI's API calls were still quite expensive for GPT4, and the cost of running it on a book-length PDF would just be prohibitive. In contrast, you could run Llama2 locally on a machine with just a CPU, and it would be extremely slow, but "free" if you had a spare machine lying around.<p>Well, it's amazing how things have changed since then. Not only have models gotten a lot better, but the latest "low tier" offerings from OpenAI (GPT4o-mini) and Anthropic (Claude3-Haiku) are incredibly cheap and incredibly fast. So cheap and fast, in fact, that you can now break the document up into little chunks and submit them to the API concurrently (where each chunk can go through a multi-stage process, in which the output of the first stage is passed into another prompt for the next stage) and assemble it all in a shockingly short amount of time, and for basically a rounding error in terms of cost.<p>My original project had all sorts of complex stuff for detecting hallucinations and incorrect, spurious additions to the text (like "Here is the corrected text" preambles). But the newer models are already good enough to eliminate most of that stuff. And you can get very impressive results with the multi-stage approach. In this case, the first pass asks it to correct OCR errors and to remove line breaks in the middle of a word and things like that. The next stage takes that as the input and asks the model to do things like reformat the text using markdown, to suppress page numbers and repeated page headers, etc. Anyway, I think the samples (which take less than 1-2 minutes to generate) show the power of the approach:<p>Original PDF:
<a href="https://github.com/Dicklesworthstone/llm_aided_ocr/blob/main/160301289-Warren-Buffett-Katharine-Graham-Letter.pdf">https://github.com/Dicklesworthstone/llm_aided_ocr/blob/main...</a><p>Raw OCR Output:
<a href="https://github.com/Dicklesworthstone/llm_aided_ocr/blob/main/160301289-Warren-Buffett-Katharine-Graham-Letter__raw_ocr_output.txt">https://github.com/Dicklesworthstone/llm_aided_ocr/blob/main...</a><p>LLM-Corrected Markdown Output:
<a href="https://github.com/Dicklesworthstone/llm_aided_ocr/blob/main/160301289-Warren-Buffett-Katharine-Graham-Letter_llm_corrected.md">https://github.com/Dicklesworthstone/llm_aided_ocr/blob/main...</a><p>One interesting thing I found was that almost all my attempts to fix/improve things using "classical" methods like regex and other rule based things made everything worse and more brittle, and the real improvements came from adjusting the prompts to make things clearer for the model, and not asking the model to do too much in a single pass (like fixing OCR mistakes AND converting to markdown format).<p>Anyway, this project is very handy if you have some old scanned books you want to read from Archive.org or Google Books on a Kindle or other ereader device and want things to be re-flowable and clear. It's still not perfect, but I bet within the next year the models will improve even more that it will get closer to 100%. Hope you like it!<p>[0] <a href="https://news.ycombinator.com/item?id=36976333">https://news.ycombinator.com/item?id=36976333</a>