A GitHub analysis of 255,068 newly written agent skills found that the share written in non-English languages rose from 13.0% in Q1 2026 to 16.3% in Q2 2026, a three-point jump in three months, compar…
Chinese-linked threat actors used open-source AI agent frameworks Hermes and OpenClaw to breach 21 Taiwanese government systems, including the nuclear safety regulator, in 12 waves of autonomous attac…
Dromeas, a code review tool, compared its LLM council-based review against Claude Code's ultrareview on a large, real pull request from the open-source project openclaw. Dromeas's three-model council …
NVIDIA released SkillEvaluator, an open-source tool that measures how AI agent skills affect performance, and published benchmark results for more than 300 verified skills across over 30 NVIDIA produc…
In a multi-repo project using OpenCode with the big-pickle model, an AI agent and its human operator developed a five-stage procedure for generating AGENTS.md and SKILL.md files, culminating in an aut…
Researchers introduced SkillConsist, a method that detects inconsistencies in LLM agent skills by separating declaration and implementation content into behavior graphs and using bidirectional graph a…
NomaDamas released k-skill, an MIT-licensed collection of 100+ agent skills for Claude Code, Codex, and OpenCode that enable AI agents to interact with Korean-specific services such as SRT/KTX train b…
A developer known as @1kalin has released the 'business-automation-architect' skill for AI agents, enabling them to automate business processes using cron jobs, scripts, and built-in reasoning without…
AgentPrizm launched on July 9, 2026 with a governed memory layer for AI agents, addressing the problem of context window amnesia via REST and MCP interfaces. The platform includes AgentMemory for pers…
Fulcra Dynamics released the fulcra-onboarding skill on ClawHub, enabling OpenClaw agents to access persistent, user-permissioned real-world context. The skill supports over 200 data sources and inclu…
Snyk's Evo Agentic Development Security platform went GA on June 29, revealing that one in 12 developers with active MCP server connections has a high or critical security finding. Scanning nearly 10,…
OpenClaw's AI agent Fiu blocked all 6,000 prompt injection attempts from over 2,000 attackers during a public test at hackmyclaw.com, demonstrating resilience against AI social engineering. The test, …
Snyk security researchers found that 13.4% of 3,984 AI agent skills from ClawHub and skills.sh contain critical security flaws, with 36.82% having at least one security issue, including malware, crede…
Unit 42 researchers discovered a malicious campaign between February and May 2026 targeting OpenClaw's ClawHub marketplace, where attackers published skills like 'tradingview-ai-indicator-assistant' t…
Four projects share the name 'SkillsGuard', but only one—AgentGuard by GoPlus Security—is a verifiable, installable runtime guard. A scanner called 'SkillGuard' was flagged as malicious and pulled fro…
Palo Alto Networks Unit 42 researchers discovered five malicious skills on OpenClaw's ClawHub marketplace between February and May 2026, including infostealers, evasion techniques, and novel agentic t…
Security firm AIR created a fake AI agent skill that bypassed every major security scanner and reached approximately 26,000 agents, including some on corporate accounts. The skill contained no malicio…
Twenty-three code-executing plugins on ClawHub, an AI agent registry, were found squatting under official @openclaw and @clawhub scopes because those scopes were not reserved for their owners, exposin…
A new study finds 78.7% of repositories using AI coding tools configure only one tool, yet developers use an average of two to four. The AGENTS.md standard, intended to bridge tools, is adopted by onl…
AgentNotes launched a new SDK that lets developers add plain-English summaries of AI agent activity to any stack with a single package install. The tool works across Python, Node.js, and OpenClaw envi…