#python #ai #llm_evaluation #llm_security #security_scanners #vulnerability_assessment

`garak` is a free tool that helps check if large language models (LLMs) have weaknesses or can be made to fail in unwanted ways. It tests for issues like hallucinations, data leaks, prompt injections, misinformation, and more. You can use it like `nmap` but for LLMs. To use `garak`, you install it with `pip` and specify the LLM model you want to test. It runs various probes to see if the model behaves incorrectly and gives you detailed reports on any vulnerabilities found. This helps ensure your LLMs are safe and reliable. You can get started by following the user guide and joining their Discord community for support.

https://github.com/NVIDIA/garak

#other #chatbot #hugging_face #llm #llm_local #llm_prompting #llm_security #llmops #machine_learning #open_ai #pathway #rag #real_time #retrieval_augmented_generation #vector_database #vector_index

Pathway's AI Pipelines help you quickly create and deploy AI applications with high accuracy. These pipelines use the latest knowledge from your data sources and offer ready-to-deploy templates for large language models. You can test these apps on your own machine and deploy them on cloud services like GCP, AWS, or Azure, or on-premises. The apps connect to various data sources such as file systems, Google Drive, and databases, and they include built-in data indexing for efficient searches. This makes it easy to extract and organize data from documents in real-time, reducing the need for separate infrastructure setups. This simplifies the process of building and maintaining AI applications, saving you time and effort.

https://github.com/pathwaycom/llm-app

#python #agent_security #ai_agents #ai_security #benchmark #claude #claude_code #codex #cursor #llm_security #mcp #model_context_protocol #prompt_injection #threat_detection

ADR is an enterprise security system for AI agents that watches what agents do, tests them for attacks, detects risky behavior, and is being used in production at Uber; its paper was accepted to MLSys 2026. The benefit to you is better protection for AI tools like Cursor, Claude Code, and Codex, plus a way to find unsafe actions before they cause harm.

https://github.com/uber/ADR