Guardrails
NeMo Guardrails is an open-source library for applying programmable safety and security controls to LLM and agentic applications.
- 7.5k
- Python
- 14 h ago
NeMo Guardrails is an open-source library for applying programmable safety and security controls to LLM and agentic applications.
A fast Rust safety hook that blocks destructive shell, git, and filesystem commands before AI coding agents run them. 50+ security packs (databases, Kubernetes, Docker, cloud, Terraform), heredoc/inline-script scanning, CI scan mode, and explain mode; works with Claude Code, Codex, Gemini, Copilot, Cursor, and more.
A reading list for large models safety, security, and privacy (including Awesome LLM Security, Safety, etc.).
Safe RLHF: Constrained Value Alignment via Safe Reinforcement Learning from Human Feedback
☕ Latte: the safest & truly intuitive templates for PHP. Engine for those who want the most secure PHP sites.
ClawKeeper: Comprehensive Safety Protection for OpenClaw Agents Through Skills, Plugins, and Watchers (aka The Norton for OpenClaw)
PyBullet CartPole and Quadrotor environments—with CasADi symbolic a priori dynamics—for learning-based control and RL
The repository is for safe reinforcement learning baselines.
🌍 A repository of blocklists for various DNS servers, featuring over 100 links and more than 6 million domains. Additionally, you can use our personalized Blocklist Generator to customize content filtering according to your individual needs, giving you full control over what should be blocked on your network.
📝 Generating, validating and processing secure forms in PHP. Handy API, fully customizable, server & client side validation and mature design.
AI jailbreaks: 35 paste-ready AI jailbreak prompts for 25+ frontier and open-weight models - system prompts, AGENTS.md specs, and Ollama Modelfiles. Built for AI security research and red-team testing.
Copy links from the sharing menu with automatic removal of shorteners and trackers.
Text classification model, used with transformers.
Gated access
🔐 NewPass is a free and open source password manager which will allow you to generate and store your passwords securely, saving them locally and encrypting them on your phone's memory
A "panic button" app for triggering a "ripple effect" across apps that are set up to respond to panic events
A survey on harmful fine-tuning attack for large language model (ACM CSUR)
[AAAI'25 (Oral)] Jailbreaking Large Vision-language Models via Typographic Visual Prompts
[ICML 2026 & ICLR 2026 AIWILD] Official Implementation of the CKA-Agent, "The Trojan Knowledge: Bypassing Commercial LLM Guardrails via Harmless Prompt Weaving and Adaptive Tree Search".
A toolbox for benchmarking trustworthiness of multimodal large language models (MultiTrust, NeurIPS 2024 Track Datasets and Benchmarks)
Code for ACL 2024 paper "TruthX: Alleviating Hallucinations by Editing Large Language Models in Truthful Space"
Claude Code-style declarative permission rules for DeepSeek Harness: ordered allow/deny/ask rules with tool-name, argument (glob/regex), and workspace-path matching on the tools/pre-execute waterfall, session-log audit, and HMR reload.
Awesome-Backdoor-on-LMMs is a collection of state-of-the-art, novel, exciting backdoor methods on LMMs (VLPs, TDMs, VLMs, and Agents).
JAX toolbox for safe robotics control: Control Barrier Function (CBF) safety filters, MPPI planning, and a drop-in safe-RL layer.
[ICLR 2024]Data for "Multilingual Jailbreak Challenges in Large Language Models"
[ICLR 2025] Robust Gymnasium: A Unified Modular Benchmark for Robust Reinforcement Learning.
[ICML 2024] Safety Fine-Tuning at (Almost) No Cost: A Baseline for Vision Large Language Models.
Offline group-safety mesh for hikers: ESP32 + GPS + LoRa, with a firmware-in-the-loop simulator and 3D replay viewer
Two-person rule CLI for AI coding agents: peer review and approval required before running potentially destructive commands
ChineseHarm-Bench: A Chinese Harmful Content Detection Benchmark
A toolbox for benchmarking Multimodal LLM Agents trustworthiness across truthfulness, controllability, safety and privacy dimensions through 34 interactive tasks
This is the official code for the paper "Virus: Harmful Fine-tuning Attack for Large Language Models Bypassing Guardrail Moderation"
Text generation model, used with gguf.
FORGE Safety - Injection + Circuit Breaker
Runs in the browser on Hugging Face
繁體中文 LLM guardrail 前處理層(偵測 + 分類 + 多輪追蹤)
Runs in the browser on Hugging Face