All published AI news and analysis.

Ruvnet has released version 3.32.4 of Ruflo, an agent meta-harness designed to coordinate autonomous AI agents. This update enhances the framework's capabilities for building intelligent multi-agent systems, offering features like adaptive memory, self-learning, and RAG integration.

The GSD Skill Creator project has released version 1.49.561, introducing eight composable modules that extend its agent-based framework for guided software development. This update focuses on enhancing AI coding assistants by addressing context rot, amnesia between sessions, and scaling complexity.

vstorm-co has released a full-stack AI agent template generator that scaffolds production-ready FastAPI and Next.js applications. The template supports multiple AI agent frameworks, RAG pipelines, real-time streaming, and enterprise-grade authentication.

AdminForth has released version 3.9.1, a framework for building agentic admin panels using TypeScript and Vue.js. The framework allows developers to set up main CRUD pages rapidly and offers extensibility for custom features.

Agent Network is an open-source framework enabling multiple AI agents, including Claude, Codex, and Grok, to collaborate and delegate tasks. It features a web dashboard for task management and supports over 8 LLM providers, running locally with data control.

LocateAnything is a new framework for vision-language grounding and detection that uses Parallel Box Decoding (PBD) to improve both speed and accuracy. Unlike traditional methods that decode 2D boxes token by token, PBD decodes geometric elements as atomic units in a single step, enhancing parallelism and preserving geometric coherence. The framework is supported by LocateAnything-Data, a large dataset with over 138 million training samples.

Researchers have introduced SpatialBench, a new benchmark designed to holistically assess the generalization capabilities of spatial foundation models across diverse tasks, viewpoints, and scene domains. The benchmark includes 19 datasets and 546 scenes, evaluating 41 models across 6 paradigms under varying input densities.

Canonry is an open-source, self-hosted platform designed for monitoring and operating Answer Engine Optimization (AEO). It offers a command-line interface (CLI), API, and web UI to track AI visibility, analyze search and traffic changes, and measure the impact of agent actions.

NocoBase has released version 2.1.37 of its open-source AI-powered no-code platform for building business systems. This update enhances collaboration between AI coding agents and human developers through a WYSIWYG interface, and integrates AI employees directly into business workflows.

Google has released version 2.0 of its open-source Agent Development Kit (ADK) for Python. This code-first framework simplifies building, evaluating, and deploying sophisticated AI agents with enhanced flexibility and control.