Urlx – an agent-made Rust replacement for curl/libcurl
Recently several AI labs have published experiments where they tried to get AI coding agents to complete large software projects. - Cursor attempted to make a browser from scratch: https://cursor.com/blog/scaling-agents - Anthropic attempted to make a C Compiler: https://www.anthropic.com/engineering/building-c-compiler A few weeks ago I posted xmloxide, an agent-made Rust replacement for libxml2 made by pointing Claude code at the libxml2 test suite: https://news.ycombinator.com/item?id=47201816 curl is arguably the most widely deployed…
What it does
In the maker’s words, at launch
Recently several AI labs have published experiments where they tried to get AI coding agents to complete large software projects. - Cursor attempted to make a browser from scratch: https://cursor.com/blog/scaling-agents - Anthropic attempted to make a C Compiler: https://www.anthropic.com/engineering/building-c-compiler A few weeks ago I posted xmloxide, an agent-made Rust replacement for libxml2 made by pointing Claude code at the libxml2 test suite: https://news.ycombinator.com/item?id=47201816 curl is arguably the most widely deployed software in the world. It's 180,000 lines of C with a history of memory safety CVEs. The curl project itself tried to address this by adding a rust-based HTTP backend using hyper, but after four years of work, they abandoned the effort in December 2024. Daniel Stenberg wrote that, "there simply were no users asking for it and there were almost no developers interested or knowledgeable enough to work on it". - Daniel Stenberg's blog post on dropping hyper: https://daniel.haxx.se/blog/2024/12/21/dropping-hyper/ - Hacker News discussion: https://news.ycombinator.com/item?id=42478911 The key insight from my xmloxide work was that an agent can iterate very quickly when given a comprehensive test suite as a specification. The majority of the development for urlx was done in two ~16 hour long iteration loops where claude code would implement some features, run the tests, and then plan what to do next. Towards then end I took a more manual approach telling it to analyze the remaining failures, make issues, and submit them to teams of subagents to fix. urlx now passes 1,300 of 1,300 applicable curl tests (100% pass rate). It supports HTTP/1.0-1.1, HTTP/2, HTTP/3 (QUIC), FTP/FTPS, SFTP/SCP, WebSocket, SMTP, IMAP, POP3, MQTT, DICT, TFTP, Gopher — with 261 CLI flags, authentication (Basic, Digest, NTLMv2, SCRAM-SHA-256, AWS SigV4), cookies, HSTS, proxy support (HTTP/SOCKS4/5/HTTPS tunnel), and a libcurl-compatible C ABI. There are some features still left to do that I might come back to. - GitHub: https://github.com/jonwiggins/urlx - crates.io: https://crates.io/crates/urlx-cli - Install: cargo install urlx-cli or brew install jonwiggins/tap/urlx The same lesson from xmloxide applies here: when a project has a good test suite, an agent can use it as a specification and converge on a working implementation surprisingly fast. curl's test suite turned out to be an excellent spec, the hard part was already done by the curl contributors over 25 years. I don't expect anyone to replace curl with this, but I think there's something worth paying attention to in the pattern: take a well-tested project, point an agent at the test suite, and get a fast reimplementation. I think this also points to a need to adopt test-driven-development when using AI coding tools, as it allows them to iterate autonomously before needing human input. I also think this changes the economics of software development, what used to be a multi-year, multi-engineer commitment is becoming a task you can meaningfully attempt in days by setting up an agent to rapidly iterate on it alone.
Does the same job
all alternatives →
- ECembedcurl.com - Pretty curl commands for your web site or blog2013 · ▲32
Hey all, As a CEO of a growing startup I don't get to code as often as I like and when I do, I try not to annoy my team with my terrible code :) So to scratch the itch, I've built some of side projects related to what our company does. The latest is embedcurl.com I see a lot of API docs and blogs posting curl commands that are really hard to read and I figured I could make them easier to read and add some niceties like easier copy/paste, tooltips on flags, etc. They also gracefully degrade if javascript isn't enabled back to a element. The parsing is done in Python with shlex and…
- ABA better way to handoff web bugs to AI agentsNov 2025 · github.com · ▲5
Hi HN, Zidan here. I’ve been experimenting with AI-assisted debugging and noticed a recurring gap: most tools optimize for agent-led exploration (ex: giving claude code a browser to click around and try to reproduce an issue). But in many cases, I've already found the bug myself. What I actually want is a way to hand the agent the exact context I just saw - without retyping steps, copying logs, or hoping it can reproduce the behavior. So we built FlowLens, an open-source MCP server + Chrome extension that captures browser context and lets coding agents inspect it as structured, queryable…
- LCLetting Claude automate fleets of browser sandboxesMar 2026 · twitter.com · ▲6
Hey HN, wanted to share this cli and skill I built @ Steel (I'm the founder) I was trying to set up openclaw on railway and quickly bumped up against issues with a lack of browser access, a core component of the system. Agents like OpenClaw and CC are super good at using browsers but, similar to code sandboxes, they should be able to control these browser in the background, in parallel, without getting blocked by captchas. So I reworked agent-browser, the most popular cli for browser agents, to navigate Steel cloud browser sessions to they can run at scale and unhindered. It's a relatively…
- AAAgentic AI Frameworks on AWS (LangGraph,Strands,CrewAI,Arize,Mem0)2025 · github.com · ▲7
We’ve published a set of open-source reference implementations on how to build production-grade Agentic AI applications on AWS. What’s in the repo: • Agentic RAG, memory, and planning workflows with LangGraph & CrewAI • Strands-based flows with observability using OTEL & Arize • Evaluation with LLM-as-judge and cost/performance regressions • Built with Bedrock, S3, Step Functions, and more GitHub: https://github.com/aws-samples/sample-agentic-frameworks-on-... Would love your thoughts — feedback, issues, and stars welcome!
More ai this month
the category →
I trained a 125M-parameter transformer to autocomplete piano performances in real time (~108 notes/sec on an iPhone 15). The idea is basically GitHub Copilot or Tabnine, except instead of prompting it with code, you prompt it by playing a few notes on a MIDI piano. The model then continues what you played, entirely on-device. The app is free if anyone wants to try it. Happy to answer questions about the model, training, Core ML, or the many things that didn't work.
AI · 17d ago · simedw.com
Astute▲585Automate your B2B brand going viral, with new media creators
AI · 18d ago · company-app.joinastute.com
Hey HN, Henry from Cactus here! We previously released Cactus Needle, a 14MB agentic LLM for tool call, device use, and structured extraction for phones, wearables, smart homes, small robots and microcontrollers. We got really great feedback here, and have now incorporated the suggestions to release Needle 2. The whole model is a single 14MB binary that runs a full session in 28MB of RAM; 45m parameters at 2bit compression. Needle hits 500 tokens/sec decode speed on a Raspberry Pi 5, sits between 400-1,500 tokens/sec on VR devices like Meta Quest 3S and Apple Vision Pro, and ranges…
AI · 27d ago · cactuscompute.com
Launched alongside, March 2026
the whole month →
Switch from ChatGPT to Claude with import memory feature
AI · Mar 2026 · claude.com
