Show HN: Open-Source AI Workflows with Read-Only Auth Scopes
Seer is an open‑source, visual workflow builder that combines drag‑and‑drop design with AI‑assisted development. It offers fine‑grained, read‑only authentication scopes and integrates with services like Google Workspace, GitHub, and web search APIs. Users can launch a full development stack via a single Docker Compose command or deploy to Railway for a managed cloud instance costing roughly $15‑30 per month. The platform separates workflow and agent APIs, enabling deterministic execution while supporting dynamic conversational agents.
The Creator of Claude Code's Claude Setup
Boris, the creator of Claude Code, shared his personal development environment, highlighting that the tool works effectively straight out of the box. He emphasized that Claude Code requires little to no customization for most users, yet remains fully extensible for...
Global Software Engineering Job Postings Outlook – 2026
The 2026 global software engineering job market features 105,115 postings, with business‑software applications (20,248) and core software engineering (16,992) leading specializations. In‑demand technical skills are heavily weighted toward Python (12,575), AWS, SQL, CI/CD, and Kubernetes, while AI‑related roles total 21,477,...
Try to Take My Position: The Best Promotion Advice I Ever Got
Andrew Graham‑Yooll shares the most effective promotion advice he received: start performing the duties of the next role before you hold the title. He illustrates this with a junior engineer who proactively drafted a service‑incident reduction proposal, embodying the "take...
Show HN: OfferGridAI – Side-by-Side Comparison of Real Estate Offers From PDFs
OfferGridAI launches a SaaS tool that instantly extracts key terms from real‑estate purchase offers and generates side‑by‑side comparison grids with AI‑driven risk scores. The platform processes up to ten PDFs per property in under two minutes, turning a task that...
OpenWorkers: Self-Hosted Cloudflare Workers in Rust
OpenWorkers is an open‑source runtime that lets developers execute Cloudflare Workers‑compatible JavaScript on their own servers using Rust’s rusty_v8 V8 isolates. The platform provides familiar bindings such as KV storage, PostgreSQL, S3/R2, and full fetch/Web API support, along with built‑in...
SigNoz (YC W21, Open Source Observability Platform) Is Hiring Across Roles
SigNoz, an open‑source observability platform positioned as an alternative to DataDog and New Relic, announced hiring across multiple roles. The project has amassed more than 21,000 GitHub stars and a Slack community of over 6,000 developers, reflecting strong market interest....
Show HN: I Built a Universal Clipboard that Syncs Realtime on Multiple Devices
A new universal clipboard service lets users copy content on any device and paste it instantly on others, boasting a 0.1‑second sync latency and end‑to‑end encryption. The web‑based app works across phones, laptops, and tablets without installing native software. It...
Google Opal
Google has launched Opal, a no‑code platform that lets users create AI‑powered mini‑apps using natural language prompts. The tool combines rapid prototyping, allowing functional prototypes in minutes, with an intuitive node‑based visual editor for logic refinement. Access is gated behind...
Honey's Dieselgate: Detecting and Tricking Testers
A recent investigation uncovered that Honey’s browser shopping extension deliberately alters its behavior when it detects potential testers. The plugin honors affiliate‑network stand‑down rules for new accounts, low‑point users, blacklisted IDs, or those with affiliate‑industry cookies, but bypasses those rules...
Everything as Code: How We Manage Our Company in One Monorepo
Kasava runs its entire platform—including frontend, backend, marketing site, documentation, and external tools—in a single monorepo. By storing everything as code, a single commit updates pricing limits across the API, UI, website, and docs without coordination. The approach enables AI...
Crimson (YC X25) Is Hiring Founding Engineers in London
Crimson, a Y Combinator‑backed AI startup, is hiring its first full‑stack engineers in London to build a litigation‑focused platform that drafts pleadings, analyzes judgments, and extracts evidence instantly. The role offers a £80‑£120K salary and 0.5‑1% equity, targeting engineers with...
I Migrated to an Almost All-EU Stack and Saved 500€ per Year
An individual migrated his personal productivity stack to an almost entirely EU‑hosted suite, primarily using Proton’s expanded services, Scaleway cloud, and privacy‑focused AI tools. The new configuration costs roughly €39 per month versus €83 previously, delivering an annual saving of...
Show HN: Stop Claude Code From Forgetting Everything
Ensue has launched a memory‑network plugin for Claude Code that lets large language models retain information across sessions. The tool builds a persistent knowledge tree, enabling users to store preferences, research, and past decisions for future reference. Installation is handled...
N8n-Oidc
Self‑hosted n8n users face a steep $400‑per‑month enterprise license to enable single sign‑on, a cost dubbed the “SSO Tax.” A new open‑source project, n8n‑oidc, injects OpenID Connect support via n8n’s external hooks without any licensing fees. The solution provides standard...
Obelisk 0.32: Cancellation, WebAPI, Postgres
Obelisk 0.32 adds cooperative cancellation for workflows and activities, letting developers abort leaf tasks while preserving compensating actions. The release also opens port 5005 to HTTP traffic, offering a simple WebAPI that returns plain‑text or JSON execution data. PostgreSQL support replaces SQLite...
Langfuse (YC W23) Is Hiring in Berlin, Germany
Langfuse, an open‑source LLM engineering platform, announced hiring in Berlin to expand its backend, product, and developer‑engagement teams. The company reports strong traction with 19,719 GitHub stars, over 23 million SDK installs per month, and more than 6 million Docker pulls. Its...
ThingsBoard: Open-Source IoT Platform
ThingsBoard is an open‑source, Apache‑2.0 licensed IoT platform that offers both on‑premise installations and a managed cloud service. It provides end‑to‑end capabilities including device provisioning, telemetry collection, rule‑engine processing, and customizable real‑time dashboards. The platform supports diverse use cases such...
Build Software. Build Users
The article argues that traditional testing cannot guarantee software quality because it often ignores real user behavior. It proposes "vibe coding"—using LLM‑driven agents that embody target users to inform development from the outset. By iteratively building user agents, defining happy‑path...
How to Construct Complex Data Declaratively and Progressively?
Pydantic‑resolve introduces a declarative, progressive approach to building complex API data in Python, eliminating the need for imperative glue code. By adding resolve and post methods, developers can fetch related entities on demand and perform post‑processing steps, positioning the library...
Building an AI Agent Inside a 7-Year-Old Rails Monolith
A Director of Engineering at Mon Ami detailed how they added an AI agent to a seven‑year‑old, multi‑tenant Rails monolith using the RubyLLM gem. By defining tool‑based function calls that wrap Algolia searches and Pundit policy checks, the LLM can...
Why FedRAMP Authorization and CMMC Level 2 Are Now Table Stakes for GovCon AI
AI is becoming core to government contracting workflows, from opportunity discovery to proposal submission. As AI handles sensitive data, FedRAMP authorization and CMMC Level 2 compliance have shifted from optional to table‑stakes for vendors. Procurement Sciences invested $40 million to achieve full...
Incremental Backups of Gmail Takeouts
The author proposes an incremental backup method for Gmail Takeout mbox files. Traditional full Takeout exports produce a 5.7 GiB file that must be re‑uploaded each backup, causing storage bloat. By splitting the mbox at each “From ” header into content‑addressed...
Postgres Extension Complements Pgvector for Performance and Scale
Timescale released pgvectorscale, a Rust‑based PostgreSQL extension that extends pgvector with a StreamingDiskANN index and Statistical Binary Quantization. The extension delivers dramatically faster approximate nearest‑neighbor searches and more cost‑effective storage for large‑scale AI embeddings. In benchmarks on 50 million 768‑dimensional vectors,...
Keystone (YC S25) Is Hiring Engineer #1 to Automate Coding
Keystone, a Y Combinator‑backed startup, is hiring its first engineer to build an AI‑native error‑monitoring platform that automatically diagnoses production issues and generates code fixes. The company raised a $5.2 million seed round from prominent venture firms and industry founders. The...
Show HN: Vibium – Browser Automation for AI and Humans, by Selenium's Creator
Vibium, created by Selenium’s founder, delivers a lightweight, single‑binary solution for browser automation tailored to AI agents and developers. The Go‑based Clicker binary bundles Chrome lifecycle management, a BiDi proxy, and an MCP server, enabling large language models like Claude Code...
Microsoft Agent Framework
Microsoft released the Agent Framework, an open‑source kit for building AI agents and multi‑agent workflows in .NET and Python. The framework merges concepts from Semantic Kernel and AutoGen, adding graph‑based workflows, robust state management, and enterprise‑grade telemetry. It supports Azure...
Show HN: A Local-First, Reversible PII Scrubber for AI Workflows
Tom Jordi Ruesch announced the open‑source release of Bridge Anonymization, a TypeScript library that enables reversible, local‑first PII masking for AI translation pipelines. The tool combines fast regex patterns with a quantized ONNX NER model to detect both structured and...
Permission Systems for Enterprise That Scale
Enterprises quickly outgrow naive read‑time permission checks, which rely on multiple recursive queries that degrade performance as data and hierarchy depth grow. Shifting to Role‑Based Access Control (RBAC) pre‑computes permissions at write‑time, allowing reads to be satisfied with a single...
Show HN: Turn Raw HTML Into Production-Ready Images for Free
html2png.dev launches a free, no‑signup API that converts raw HTML into production‑ready images. Users can POST HTML directly and receive PNG, JPEG, WebP, or PDF files with customizable dimensions, retina scaling, and transparent backgrounds. The service returns an immutable URL...
Is Northern Virginia Still the Least Reliable AWS Region?
StatusGator’s 2025 analysis confirms that the US‑East‑1 (Northern Virginia) AWS region remains the least reliable, recording ten public outages, 33 hours 49 minutes of downtime, and 126 affected components—the highest across all commercial regions. Compute‑focused services such as EC2, ECS and EMR dominated...
Texas App Store Age Verification Law Blocked by Federal Judge
A Texas federal judge issued a preliminary injunction blocking the Texas App Store Accountability Act, which would have forced Apple and other marketplaces to verify every user’s age starting January 1, 2026. The judge ruled the law likely violates the First Amendment,...
HTTP Caching, a Refresher
The 2022 RFC 9111 update codifies Cache‑Control as the central mechanism for directing browsers, proxy caches, and CDNs on how to store and reuse HTTP responses. It defines freshness calculations via max‑age, s‑maxage, Expires and heuristic methods, and outlines validation using...
Meta Is Using the Linux Scheduler Designed for Valve's Steam Deck on Its Servers
Meta engineers revealed that they have adopted the SCX‑LAVD Linux scheduler, originally created for Valve’s Steam Deck handheld, as the default scheduler for their large‑scale server fleet. The scheduler, developed by Igalia, delivers latency‑aware virtual deadline scheduling and matches or exceeds...
Instant Database Clones with PostgreSQL 18
PostgreSQL 18 introduces the file_copy_method=clone setting, turning database cloning into an almost instantaneous, zero‑copy operation. By leveraging filesystem reflink capabilities on XFS, ZFS, APFS and similar, a clone can be created in milliseconds instead of minutes. The feature builds on the CREATE...
Things I Learnt About Passkeys when Building Passkeybot
Passkeybot.com launched a hosted sign‑in page that lets developers add passkey authentication with minimal server code. The author details technical insights gained, from Apple’s Secure Enclave and the distinction between user presence and verification to attestation’s privacy trade‑offs. New WebAuthn...
Uplane (YC F25) Is Hiring Founding Engineers (Full-Stack and AI)
Uplane, a Y Combinator F25 startup focused on AI-driven marketing, announced three founding‑engineer openings in San Francisco. The roles include a Founding AI Engineer and a Founding Full‑Stack Engineer, each with salaries ranging from $150,000 to $200,000, plus a Business...
Show HN: Netrinos – A Keep It Simple Mesh VPN for Small Teams
Netrinos launches a WireGuard‑based mesh VPN aimed at small and remote teams, offering automatic device discovery and firewall‑bypassing connectivity. The service works on Windows, macOS, Linux, Raspberry Pi, cloud VMs and IoT devices without any port‑forwarding or IT intervention. Pricing includes...
Reinventing the Dial-Up Modem (2019)
Simple, a healthcare app, replaced paper registers and landline calls with an offline‑first solution that uses DTMF tones to transmit patient numbers through a toll‑free line. Nurses initiate a call, the app plays the patient’s number as DTMF, and Twilio...
Show HN: Spice Cayenne – SQL Acceleration Built on Vortex
Spice.ai announced Spice Cayenne, a next‑generation data accelerator that pairs the Vortex columnar file format with an embedded SQLite‑based metadata layer. The solution targets multi‑terabyte object‑storage workloads, delivering up to 1.4× faster TPC‑H queries and 14% faster ClickBench runs while...
Show HN: Python SDK – Forecasting with Foundation Time-Series and Tabular Models
FAIM has released a production‑ready Python SDK that unifies foundation AI models for time‑series forecasting and tabular inference. The package ships with multiple state‑of‑the‑art models—including FlowState, Amazon Chronos 2.0, TiRex for series and LimiX for classification/regression—exposed through a type‑safe, Pydantic‑validated API....