Alice Launches Caterpillar to Detect Malicious OpenClaw Skills
Alice has released Caterpillar, a free open-source security tool aimed at protecting AI agents in OpenClaw, announced in a press release. The software helps developers and security teams identify unsafe or malicious behaviors in agent skills as the OpenClaw ecosystem continues to expand.
Caterpillar analyzes skill logic and configurations to detect injection paths, unsafe tool access, and obfuscated code. It draws on data from Alice’s RabbitHole adversarial intelligence platform, which compiles threat signals from real-world research.
Before its public release, Caterpillar detected several malicious skills in OpenClaw, including ones used by more than 6,000 users. The tool is now available as open-source software, allowing the community to audit and extend its capabilities for securing AI agent workflows.
We hope you enjoyed this article
Consider subscribing to one of our newsletters like Cybersecurity AI Weekly, AI Policy Brief or Daily AI Brief.
Also, consider following us on social media:
More from Cybersecurity
Oct 2 Suspected Chinese spies impersonate AI policy figures in phishing campaign Oct 2 AI Agents Tried to Access Canadian Government Archive Site Oct 2 Nuix Showcases Neo Financial Crime Tools at Sibos 2026 Oct 1 OpenAI Links Model Reasoning Extraction Campaign to Moonshot AI Oct 1 Ocrolus Adds Resistant AI Document Fraud Detection to DetectMore from AI Safety
Oct 2 OpenAI Parts Ways With Three Safety Researchers Oct 1 OpenAI Links Model Reasoning Extraction Campaign to Moonshot AI Sep 30 Chinese AI Agents Deceived Evaluators in Controlled Tests Sep 29 Florida Attorney General Seeks to Block New OpenAI Model Development Sep 29 UK Safety Test Finds GPT-6 Astra Conducted Simulated Supply Chain AttacksCybersecurity AI Weekly
Weekly newsletter about AI in Cybersecurity.
Whitepaper
Tensordyne Napier: What If One Rack Could Do the Work of Nine?
Tensordyne
This Tensordyne whitepaper presents Napier, an inference-focused AI processor and rack-scale system based on the company’s TDN Math logarithmic number system. It examines infrastructure requirements for large mixture-of-experts and agentic models, compares major inference architecture approaches, and details the TDN AIP processor, TDN72 pod, TDN Link fabric, and Napier Ultra configuration. The paper reports simulation-based performance, cost, and accuracy-validation results, including Tensordyne’s projected comparison of one Napier rack with a nine-rack Nvidia Rubin plus Groq deployment; the chip is reported as taped out and in fabrication.
Read moreYou may also like
Anthropic's Threat Report Finds AI Moving From Assistant to Orchestrator
Noma Security Extends Agent Controls to Employee Endpoints
Anthropic Attributes Its Largest Measured Distillation Campaign to Alibaba
Commvault Adds Recovery for Claude and OpenClaw Data
Researchers Link OpenAI Agents to May RubyGems Attack
Daily AI Brief: the AI news that matters, in your inbox.