Companies tracked by Rails Radar

Hiring signals with observed stack evidence. Showing 193–240 of 1375 · page 5 of 29.

Bjak Hiring 1 open listing

url=https://remoteOK.com/remote-jobs/remote-engineering-manager-thailand-bjak-1136670, title=Engineering Manager Thailand
Blackholes 0 open listings

discussion=https://news.ycombinator.com/item?id=49617072, title=Show HN: Blackholes – A macOS workspace for coding agents, Hi HN, I'm building Blackholes, an open-source macOS app for organizing coding agents, Git repositories and terminals. A project can contain multiple repositories, and tasks can have their own branches and worktrees. For example, you can work on a feature across a frontend and backend repo, keeping both under the same task. It supports Claude, Codex, Gemini and OpenCode, with project and task notes, file editing and Git diffs available in the workspace. The app is built with Rust and React. Source code: https://github.com/andygeek/blackholes I'd appreciate feedback from people working with coding agents across multiple repositories, especially on what would make this fit into your existing workflow.
Blast Arena 0 open listings

discussion=https://news.ycombinator.com/item?id=49731107, title=Show HN: Blast Arena – Quake 3-style deathmatch in the browser (WebRTC P2P)
Blender Agent:A live, multi-turn agent that any can use for Blender 0 open listings

discussion=https://news.ycombinator.com/item?id=49654524, title=Show HN: Blender Agent:A live, multi-turn agent that any can use for Blender
Blender MCP by ArchBench 0 open listings

discussion=https://news.ycombinator.com/item?id=49671951, title=Show HN: Blender MCP by ArchBench
Blender in Docker for computer-use agents 0 open listings

discussion=https://news.ycombinator.com/item?id=49622926, title=Show HN: Blender in Docker for computer-use agents
BlindLock 0 open listings

discussion=https://news.ycombinator.com/item?id=49642803, title=Show HN: BlindLock – Passwords hidden in a photo, sealed to your TPM/SE, Hi everyone, I'm David, the solo developer and owner of BlindLock. This is my second post here. BlindLock is more than a year of work. Not a weekend hobby project I threw together. I've put a hell of a lot of time, heart, and nerves into it. Why steganography? Picture a burglar. He looks for valuables, the obvious ones: gold, cash, jewelry, the safe in the basement. The picture of your kid on the fridge, the holiday photo on the wall, are irrelevant to him. That's exactly the principle BlindLock is built on. Hackers go after cookies, passwords in the browser, container files that can be decrypted with a password on any computer, crypto seeds, and other sensitive data. Anyone who can log into your Google account or your ChatGPT knows more about you than they could learn in a personal conversation with you. All of that can be grabbed with RATs and other malware, passwords included. BlindLock hides exactly these sensitive logins and passwords inside a picture, because a photo has almost no value to an attacker. And he'd first have to identify the picture as his target at all. On top of that, BlindLock seals this picture to the security chip of your computer. That means: even if an attacker copies the picture, together with the password, he cannot get at your data. The same chip sealing applies to the container vault files. And for these larger files BlindLock has a built-in file explorer, so you can open and edit files without mounting them into the OS. Wherever the respective OS allows it, I've written it to leave no traces behind. Why isn't BlindLock open source? That was the most frequent question. Take Bitwarden as an example. Bitwarden sells a service: servers, sync, accounts. The code can be open because the value isn't in the code. I sell the program itself. There's no service behind it, no cloud, nothing on my side that the customer would need. The entire value is in the program itself. That's why Bitwarden can give the code away and make it public for everyone, and I can't. What's open on my side is what has to be open: the cryptography libraries are freely available and partly formally verified (HACL, level 5, the highest). Anyone can audit them, down to the smallest detail. The network: BlindLock is offline at its core The main software never has an internet connection. Nothing goes out. That's not a setting, it's the build, the architecture: BlindLock is built so it can't reach the internet at all. There is no HTTP client in the program. There is no code, no module, no function that could open a connection to the outside. Everything that needs the server runs in a separate small program, the Net-Helper. It ships in the same package and installs with BlindLock, but runs as its own process with its own binary. BlindLock starts it, asks it a question, and gets an answer. No port, no socket, just the child process's standard input and output. The helper never sees a vault or a password. It has exactly two jobs: it checks the license, and it checks whether the latest BlindLock version is installed. If there's a new version, it downloads the update package and hands it over. In short: when you open BlindLock, it pings the server through the helper. That's it. There is no other internet connection. You can also use BlindLock offline for 7 days at any time. Tech stack, licenses, and WalletLink in my first comment below. Happy to answer everything here personally. Anyone can try BlindLock free for 7 days. No email, no payment details. David
Blue 0 open listings

discussion=https://news.ycombinator.com/item?id=49728525, title=Show HN: Blue – open-source governance for coding agents, Hey HN, After migrating to an inference gateway to serve internal usage across an organization, problems have started to pop up. Once you detach from a managed ecosystem such as OpenAI or Anthropic you lose governance capabilities that makes the transition difficult. - Losing access to a central plugin hub for delivering MCPs, skills, etc. A gateway can sort of do this but has low compatibility beyond skills, different harnesses, and version ranges - Inability to deliver managed config (model defaults, etc). Can sort of be done with MDM/Jamf but doesn't reach contractors and doesn't follow employees to other machines - Gateway configuration is awkward for devs transitioning from SSO and API keys sit in plain text on their machines The gateway transition facilitates new model families (Kimi, Deepseek, etc) which are best utilized with new harnesses beyond Codex or Claude Code. Managing these additional tools multiplies the governance problems. Lastly, attempting to deliver an attribution program which can tag inference requests with ticket numbers and other metadata to facilitate Opex/Capex calcs for finance becomes fairly unfeasible when there are multiple harnesses across arbitrary versions. This was not solved with managed ecosystems either to be fair, but requires some self-serve program or some local MITM proxy which is a distribution and versioning problem in of itself. I built Blue to attempt to solve these problems. It's an MIT open source, self-hostable control plane. Developers install a universal CLI and point it at their hosted control plane tenant. On launch, it reconciles the organization's policy per harness, delivers extensions, and configures the gateway. It then launches and passes control to the native CLI (Kimi, OpenCode, Codex, etc) that devs already use. Authentication is via browser device authorization which mints ephemeral bearer tokens. For gateway access, these tokens are exchanged at the inference proxy (part of the self-hosted Blue image) for the real API keys. This means that devs don't need to manage API keys or even log into the gateway at all. Lastly, SCIM and OIDC are supported out of the box, so orgs can connect their identity solution immediately. Curious to hear your thoughts and if you have/solved this problem. I believe it can also be solved to some degree with a harness like Pi or even OpenCode, but that of course depends on the entire org migrating to those. Last thing I'll say is nowadays a CLI isn't enough since I've seen over 80% of devs prefer desktop apps like the ChatGPT app. We're also working on an open source desktop app that supports metaharnesses like Blue to make this transition easier. Thanks!
Blur. Sharpen. Repeat. Get Turing Patterns 0 open listings

discussion=https://news.ycombinator.com/item?id=49693830, title=Show HN: Blur. Sharpen. Repeat. Get Turing Patterns
BookBeat Hiring 2 open listings

url=https://bookbeat.teamtailor.com/jobs/8202101-senior-ml-engineer-model-builder-and-trainer, title=Senior ML Engineer, Model builder and trainer
Booley 0 open listings

discussion=https://news.ycombinator.com/item?id=49622684, title=Show HN: Booley – open-source IDE for agentic chip design, I am a digital design engineer, and my day job is designing chips (mostly IP blocks, not full chips) in SystemVerilog. At the start of 2026 I started experimenting with LLM agents like Claude and Codex, and realized that they are very capable not just at software and scripts, but at hardware design as well! But after the initial excitement, I started running into repeating failures and friction: - In chip design, we run a lot of heavy EDA tools: simulation, synthesis, linters. LLM agents often skipped those checks unless explicitly prompted
Botbin 0 open listings

discussion=https://news.ycombinator.com/item?id=49636340, title=Show HN: Botbin – Pastebin for AI agent artifacts
Botbin.io 0 open listings

discussion=https://news.ycombinator.com/item?id=49639496, title=Show HN: Botbin.io – pastebin for AI agent artifacts
Bough, the agent I built to replace Claude Code at work 0 open listings

discussion=https://news.ycombinator.com/item?id=49711939, title=Show HN: Bough, the agent I built to replace Claude Code at work
Bounce Router. A TUI over Claude, Codex and Muse with Usage Failover 0 open listings

discussion=https://news.ycombinator.com/item?id=49616089, title=Show HN: Bounce Router. A TUI over Claude, Codex and Muse with Usage Failover
Bounty Index ,compare public bug bounty programs across 6 platforms 0 open listings

discussion=https://news.ycombinator.com/item?id=49628531, title=Show HN: Bounty Index ,compare public bug bounty programs across 6 platforms
Breakdown: a daily game about guessing data distributions 0 open listings

discussion=https://news.ycombinator.com/item?id=49721246, title=Show HN: Breakdown: a daily game about guessing data distributions, Breakdown is a daily game where you guess how some data is distributed. Today's puzzle is "what is the world's habitable land used for." Make your best guess and then reveal the answer. You might be surprised by how far off you are (or how close). There's a new question every day, and you can also play all past questions in the archive. No account required and no ads. Inspired by the Ipsos Perils of Perception surveys, it's also an attempt to gamify testing our knowledge and biases. If this is interesting, you should also check out Perils of Perception: https://www.ipsos.com/en/perils/perils-perception-2024
Brewser, a web engine from scratch for the modded Nintendo Switch 0 open listings

discussion=https://news.ycombinator.com/item?id=49711397, title=Show HN:Brewser, a web engine from scratch for the modded Nintendo Switch
Briff 0 open listings

discussion=https://news.ycombinator.com/item?id=49626806, title=Show HN: Briff – Site Summarizer, I read HN everyday and some posts are way too long for my short attention span, so I've been wondering about summarizing long articles using AI and as I use Brave it comes with an AI sidebar tool that has a Summarize option, which is cool but... it doesn't work in Venezuela So I built this little tool and thought I'd share with you all. It's just a summarizer, enter a URL and AI (DeepSeek) writes an executive summary in less than 100 lines plus some key bullet points. Enough to get an understanding of the topic covered so I can decide to read the whole article or not I hope you all enjoy it and please don't abuse it as it cost tokens (1 cent per summary). I may think on adding ads in the future but I hate ads, so token consumption will decide. Perhaps limiting it to 3 daily summaries per user capping at 100 per month? Time will tell Feedback is highly appreciated https:/briff.site
Brite Payments 0 open listings

url=https://careers.britepayments.com/, title=150+ (their site, 3 days ago), matching_tags=Rails, description=Stackholm stack: LLM|Rails
Browse PC Parts on Amazon.com without distractions 0 open listings

discussion=https://news.ycombinator.com/item?id=49655710, title=Show HN: Browse PC Parts on Amazon.com without distractions, I've built a website where you can browse PC parts from Amazon without distractions and clutter. Most people, like myself, value more information and a clear way to find products than being bombarded with irrelevant stuff like "hey buy this item too" or "people also buy this" Sometimes you need to scroll too far down just to see actual specs. So I decided to make it simple, just a list of products and some filters to find what I need.
Browser LLM Fit 0 open listings

discussion=https://news.ycombinator.com/item?id=49617427, title=Show HN: Browser LLM Fit, Auto-detect client hardware capabilities and match against in-browser AI models across WebGPU, WASM, and ONNX Runtime. GH: https://github.com/hemanth/browser-llm-fit
Browser only CSV editor, data never leaves your machine 0 open listings

discussion=https://news.ycombinator.com/item?id=49605314, title=Show HN: Browser only CSV editor, data never leaves your machine, The current software I am working upon as part of my full time job, I have to deal with very long JSONs and lot of CSVs. For editing or viewing JSONs my first and obivous choice was jsonher.io, it’s the most wonderful JSON viewer I have ever seen. But I wish I could the same for CSVs too, when I googled it, I couldn’t find anything good.Most of the tools were asking me to sign up, some of them allowed me to open CSV but then came a big popup asking for my credit card so that I can edit it or use some filters, and lot of them were sending CSV to server in case it was large. I will be honest, there were some which were no signup or no server, but their UX was bad - sublime was showing the content better than those.Then I started building one for myself, and it turns out building no signup, browser only CSV tool is actually possible even for managing large CSVs; I was able to make it up and running within days and it turns out to be awesome - apologies for shameless bragging.Later I added support for downloading content in JSON and other alternatives of CSV like PSV, TSV, SSV, etc. It has all the basic functionalities for working with CSVs - filtering, sorting, editing, deleting, downloading in your desired format.In near future, I am planning to make it one shot CSV editor for all kinds of platform - browser extensions, desktop app and hopefully mobile apps too ( I am bad at building mobile apps, but I will try my best ). Do check it out and let me know in comments, what you feel about it.
Browser viz of OpenAI's "spaghetti" Navier–Stokes vortex (illustration) 0 open listings

discussion=https://news.ycombinator.com/item?id=49616564, title=Show HN: Browser viz of OpenAI's "spaghetti" Navier–Stokes vortex (illustration)
Browser-based Nginx config generator with live preview 0 open listings

discussion=https://news.ycombinator.com/item?id=49619076, title=Show HN: Browser-based Nginx config generator with live preview, Generates Nginx server block configs for
Build Morocco 0 open listings

discussion=https://news.ycombinator.com/item?id=49693536, title=Show HN: Build Morocco
Build2me 0 open listings

discussion=https://news.ycombinator.com/item?id=49702895, title=Show HN: Build2me – Prove2Me for software agent swarms build code, Hi everyone, this is my first original post on Hacker News. I’d like to share my thoughts on the next generation of AI agents. I believe these agents will operate autonomously over long periods. While current chat interfaces are designed for conversation, the interfaces for next-gen agents should focus on delegation; they will be capable of gathering environmental data, interpreting requirements, and working continuously toward a goal for days or even a month. That is why I ported prove2me and built the build2me project. My inspiration came from prove2me , which uses Lean to run multi-agent systems for long-term mathematical proof verification—a process requiring extensive exploration and prolonged task execution. I believe this same logic applies to software development and long-term self-iterating projects. I invite you to try it out and please feel free to reach out with any feedback. Thank you.
Built myself a super YouTube post-prod studio 0 open listings

discussion=https://news.ycombinator.com/item?id=49611999, title=Show HN: Built myself a super YouTube post-prod studio, i just want to film myself rambling and all the cutting, grading, uploading, thumbs, short generation is done for me. 31k lines of code for now. when you're too poor to afford an editor, you pay for claude code max to build one
Bulkgrid 0 open listings

discussion=https://news.ycombinator.com/item?id=49712405, title=Show HN: Bulkgrid – Search websites and GitHub repositories from your AI agent, Bulkgrid gives AI agents searchable knowledge from sources you choose and trust. You can use existing public sources or index a website or GitHub repository. Bulkgrid will process the content, check for changes automatically and provide relevant passages with references for your AI agent. The sources can also be curated into collections if you have agents that are doing recurring work and need reliable knowledge for that. Private repositories can be connected through GitHub and stay private to your workspace. You can connect through MCP, REST API or TypeScript and Python SDKs or use our CLI. Any feedback would be welcome! Website: https://bulkgrid.com Repository: https://github.com/bulkgrid/bulkgrid
Busy Door 0 open listings

discussion=https://news.ycombinator.com/item?id=49694470, title=Show HN: Busy Door, A cool way to keep your family informed about your upcoming and ongoing meetings in your WFH setup.
Buymydeadstartup.com 0 open listings

discussion=https://news.ycombinator.com/item?id=49625537, title=Show HN: Buymydeadstartup.com – Sell your IP to AI labs, I have had my fair share of failed startups, but none so recent that I could have sold our IP (code, Slack messages -- you name it) to Anthropic, OpenAI or any other AI lab or data buyer. I know some companies are actively approached by some brokers, but most are not, because a startup often dies as a zombie (= not in a big bang, in most cases). BuyMyDeadStartup is a bulletin board, not a marketplace. A founder posts a text description of what the company left behind, verified against a working email address and read by a person before it goes up. Buyers write to the founder directly. Whatever they agree is between them. I really tried to keep this as simple as possible.
Bypassing Transformer Softmax via Static Contraction 0 open listings

discussion=https://news.ycombinator.com/item?id=49666335, title=Show HN: Bypassing Transformer Softmax via Static Contraction
CUDA/graphics in QEMU-KVM VMs without passing the Nvidia card to them 0 open listings

discussion=https://news.ycombinator.com/item?id=49609389, title=Show HN: CUDA/graphics in QEMU-KVM VMs without passing the Nvidia card to them, Hi HN. I built this because I wanted a VM with a real GeForce GPU without losing the card on my desktop.It allows you to run CUDA and graphics in VMs without VFIO/vGPU by forwarding the NVIDIA driver's own ioctl surface from the guest to the host.That means stock libcuda and real Vulkan ICD run in the guest - no CUDA/Vulkan API remoting or translation. Display, LLM, CUDA, headless supported. Your desktop keeps using the same card, multiple VMs supported.Happy to go into the details of the architecture.
Cadenya 0 open listings

discussion=https://news.ycombinator.com/item?id=49651207, title=Show HN: Cadenya – An Agent Runtime, My name is Robert, and I'm the founder of Cadenya, an agent runtime. Cadenya is not like other managed solutions. Instead, it uses your API or MCP servers as the tools in your agentic loop, which also is managed. Cadenya is also agnostic to the model provider. Wire up OpenRouter, or any OpenAI compatible endpoint, and it will use it for inference. It also supports agent variations - where Cadenya will select the agent's with the best feedback using Thompson Sampling. And finally, Cadenya can be used in the frontend. Widgets receive their own hostname with CORS settings, and you can leverage the widgets-sdk and react components to build an agentic experience in the browser, without any complex backend logic. Email [email protected] with the Subject "Saw on HN" and get a promocode that unlocks a free account with $20 of credit.
Caffenol, a darkroom simulation photo editing toy 0 open listings

discussion=https://news.ycombinator.com/item?id=49663827, title=Show HN: Caffenol, a darkroom simulation photo editing toy, Caffenol is a darkroom toy, built on film development and darkroom craft. You load a photo and develop it into a negative in a simulated tank. Then you print that negative under a timed enlarger exposure. You shape the light by hand with cut cards that you wave over the paper. Under the hood, it's a WebGPU-powered film and paper sim, with chaos from each run producing a unique, organic result. It's built to encourage experimentation and there are few guardrails. I built this because darkroom printing made me fall in love with editing photography in a way Lightroom or Capture One couldn't. But it's expensive, uses a lot of resources, and isn't accessible. So, Caffenol brings the process and makes a few opinionated turns so it's fun and enjoyable. Ultimately, I hope it helps people make cool art. I'm tuning the film and paper sims with B&W processes now, with color coming shortly (read: when the color process can be made as fun as B&W is). No signups and everything is stored locally. If you find yourself wanting to learn more about how to bend and break it, the manual might help: https://caffenol.app/manual/ Hope it can make a print worth saving.
Calmscroll 0 open listings

discussion=https://news.ycombinator.com/item?id=49644043, title=Show HN: Calmscroll – a reader that shows the current paragraph in the library
CanWeShip 0 open listings

discussion=https://news.ycombinator.com/item?id=49673193, title=Show HN: CanWeShip – will USPS, UPS or FedEx accept my item?
Candle 0 open listings

discussion=https://news.ycombinator.com/item?id=49694897, title=Show HN: Candle
CantonPay 0 open listings

discussion=https://news.ycombinator.com/item?id=49635070, title=Show HN: CantonPay – registration-free Stellar payment links, I built CantonPay because I wanted a simple way to accept Stellar payments without creating a merchant account. You can create a payment link without registering, send it to someone, and receive the payment directly to your Stellar wallet. It also has an API and webhooks for integrating payments into an application. The current version is running on Stellar Testnet, so there’s no real money involved. I’d especially like feedback from developers who have built payment flows on Stellar.
Capsule 0 open listings

discussion=https://news.ycombinator.com/item?id=49712278, title=Show HN: Capsule – Single-file web apps that save their data into SQLite, Hey HN, I always had the problem that building HTML pages is really simple now, but trying to save data required hosting it somewhere, and sharing it afterwards was not easy. Over the last few months, I've been building an app called Capsule (it’s also the file extension name) written in Rust with Tauri 2.0 that allows packing an HTML app and its data into a single SQLite file. The HTML file and any related assets are directly embedded in the database. User data can either be saved as a localStorage key/value store or via a MongoDB-inspired collections API as documents, saved in a table in the file. You can also save other assets, like PDF files or images, directly in the database to keep different documents together. All data can be easily exported to CSV or JSON if needed. Privacy and security were a big priority for me, so documents cannot do anything out of the box. They don’t have direct access to the file system and they require permission to access the internet. The permission model is still something I’m working to improve. Capsule documents can also use local or remote AI models for document specific AI features. One downside with this approach is that multiple people working on it will create different copies. To make it possible to merge different copies of the same file, each data entry has a unique UUID and timestamp. I’m planning to open up the file format specification for the 1.0 version of the app so other apps can read or write Capsule files. You can try it out in the web preview at https://withcapsule.app/preview with pre-built templates or use any AI provider of your choice to create a custom, Capsule-optimized app by using the following prompt: "Please read the app wizard instructions at https://withcapsule.app/prompt.txt and help me design an app.“ I’m still working on the file format but there are migrations for each new version, so data should never be lost when using newer versions of the app in the future. Please let me know if you have any ideas or use cases where this might make sense or does not work.
Caribou 1 open listing
Castforge 0 open listings

discussion=https://news.ycombinator.com/item?id=49626524, title=Show HN: Castforge – run Claude Code, Codex and Gemini as one dev team
Centio Consulting Group AB 0 open listings

url=https://arbetsformedlingen.se/platsbanken/annonser/31469293, title=Applikationsexperter till Strömmande dataplattform, description=...tveckling i .NET. Minst 1 års arbetslivserfarenhet av scriptning i Python, PHP, Ruby eller PowerShell. Övriga krav Säkerhetsprövning: Säkerhetsprövning och registerkontroll krävs. Om oss Centios övergripande mål är att vara ett konsultbolag...
CertArena 0 open listings

discussion=https://news.ycombinator.com/item?id=49653516, title=Show HN: CertArena – Free AI certification simulator (zero deps, native Node 24)
ChaosTree 0 open listings

discussion=https://news.ycombinator.com/item?id=49694404, title=Show HN: ChaosTree – A zero-dependency Java tree library (AVL,RBT,B-Tree,B+Tree), What is ChaosTree? ChaosTree is a zero-dependency Java Sorted Set/Map library built around multiple tree implementations. It provides implementations of: - AVL Tree - Red-Black Tree - B-Tree - B+Tree I have not made my custom API as it implements NavigableSet, NavigableMap, SequencedSet and SequencedMap. My Custome API are: - buildFromSorted(Iterator> it, float factor) - importFlatMatrix(Object[][] blast, float factor) - Object[][] exportFlatMatrix() I started ChaosTree because I wanted to explore how far I could learn DS structure, It started as frst release of 7 tree of Set type with Custome API, As i got engrossed into it I moved to Map, During phase of switching from Set to Map there was drastic change and knowledge upgrade where I pushed my Tree limit from jdk11+ support to jdk21+ where paid close attention to dependency-free and memory layout, allocation,JVM behavior, and real-world performance. Some of the things I experimented with: - Different node layouts and metadata footprints - CRTP/F-bounded polymorphism for tree implementations - Parent-pointer vs parentless nodes - Array-based N-ary tree nodes - B-Tree/B+Tree degree selection - JMH benchmarking and JFR profiling - Differential/randomized testing against java.util.TreeMap/TreeSet After this rough and tough It also passes these test: -Guava Testlib compatibility testing -jqwik property-based testing -Randomized differential testing against reference collections -White-box structural validation of tree nodes -Direct validation of B-Tree/B+Tree structural invariants -Exception and iterator-contract testing -Serialization and cloning tests -Nary Tree uses custom jqwick test of API validation as well. Tail latency behavior is not shown here because it get's truncated into a simple text which made the dat read wrong: https://chaos-vy.github.io/ChaosTree/utils/JMH-Report.html I have also ran benchmark with official JDK TreeMapUpdate to my N-ary tree. GitHub: https://github.com/Chaos-vy/ChaosTree https://chaos-vy.github.io/ChaosTree/ I'd especially like feedback on the API design, implementation choices, benchmark methodology. I am currently trying to truncate useless and complex branches for performnace tuning.
Charisma-tec Hiring 1 open listing

url=https://weworkremotely.com/remote-jobs/charisma-tec-junior-devops-engineer-ci-cd-developer-tooling-remote-m-f-d, title=Junior DevOps Engineer (CI/CD & Developer Tooling) remote (m/f/d)
Charter 0 open listings

discussion=https://news.ycombinator.com/item?id=49649759, title=Show HN: Charter – Operate production-safe agents that run on your own infra, Even though everyone is talking about AI agents doing everything for them at work, many companies still aren't using autonomous agents to automate real operational work internally. Usually it's because of the complexity of bootstrapping an agent from scratch that’s production-safe (won’t spend all your company’s budget, burn through compute costs because it runs too often, or call a tool that breaks a customer). Even with today's agent-building frameworks, running agents reliably in production often means stitching together LangChain, Temporal, observability tools, approval systems, and custom guardrail logic. And that’s just for running an agent safely: there also has to be a management system so someone can observe and operate all their agents (upgrade it, roll it back, stop it). We wanted to build the end-to-end internal agent infrastructure that every company has to build today to use production-safe agents internally as an open-source platform. The important thing is that getting an agent running safely is quick and simple: someone can first define their agent tying it to a version, and their lifecycle and runtime policies in YAML. Next they register their agents on workers in YAML and run those workers wherever they want. This means that if one AKS cluster has permissions for the tools needed for agents X and Y and another has permissions for agents A and B, they would each register those agents on their respective clusters. Finally they can use the CLI to apply and run the agent, and then manage it with approvals, responding to questions its asking, upgrade it/roll it back/pause it (via either CLI or a UI/portal running on localhost that are communicating with the control plane). Some of the agent lifecycle management can be automated by the system: metric-based lifecycle policies make it so a user can say “if this new agent version I’m trying out (with a different prompt) fails 10 times, roll it back to the old version” or “if the new agent version is spending over a certain amount in the last 5 runs block it from being run anymore till I debug and resume it”. This is useful for teams that want to run many agents without needing someone on standby to intervene if something goes wrong on any one of them. The control plane is self-hostable or can be run via a managed cloud offering, and the self-hostable option combined with someone doing local inference is actually a completely air-gapped system. Give it a try: https://github.com/boundflow/charter and let us know what you think.
Check an NPM package or MCP server for malicious code before install 0 open listings

discussion=https://news.ycombinator.com/item?id=49722560, title=Show HN: Check an NPM package or MCP server for malicious code before install