Articles, releases and code from Hacker News, Reddit, GitHub and the people building RPA, workflow automation and AI agents — plus what the community pushed to the top today.
AIMistral launched Agentic Search, a multi-step retrieval toolkit that lets AI agents iteratively search, open, and navigate complex internal documents with lower latency and higher accuracy.
AIRubato is an open-source ESP8266 desk device that tracks AI coding agent status over MQTT to trigger physical rest breaks during long generation tasks.
AIAria Icons released an MCP server and CLI allowing AI coding agents to search, retrieve, and inject over 340,000 SVG icons directly into projects.
AIRobin launched a browser extension that lets users restyle websites, automate repetitive browser tasks, and connect web pages to external tools using natural language prompts.
AIRouted released an open-source, local hybrid router that dispatches AI agent skills using CPU-based search, eliminating LLM routing costs and network latency for local agent workflows.
AIA public, human-monitored noticeboard enables AI agents to post messages, coordinate with other models, and pass persistent notes across instances via simple HTTP requests.
AIThe PV-PP Runtime API entered public beta, giving developers a standardized decision-governance framework to evaluate and license AI agent actions separately from host application state.
AIBrowser-bridge provides governed credential autofill for browser-driving AI agents, allowing them to log into websites without exposing raw passwords to the model.
AIAAFP Commons provides an immutable, policy-gated local signed ledger with MCP and CLI interfaces, allowing AI agents to record cryptographically verifiable claims and provenance offline.
AIAgentkind launched a network that gives autonomous AI agents persistent identities, shared memory storage, and communication relays across local and global environments.
AIByteByteGo breaks down how Model Context Protocol, RAG, and AI agents differ, helping engineers choose the right architecture for tool integration, data retrieval, and autonomous workflows.
AIOpenAI launched GPT-6 Astra for developers, enabling automated workflows to generate detailed 3D models and complex outputs through improved prompt comprehension.
AIFigma built specialized AI agents to triage security alerts and draft code fixes, demonstrating how compartmentalized memory and human-in-the-loop workflows accelerate incident response safely.
AIGoogle open-sourced Mantis, a multi-agent framework that automates software vulnerability discovery, sandboxed reproduction, and patching while reducing false positives in security pipelines.