guardrail 开源项目推荐（100 个）

datreeio /datree

再次防止 Kubernetes 配置错误进入生产环境！从代码到云端，Datree 提供端到端策略执行解决方案，自动检查规则违规情况。请参阅我们的文档：https://hub.datree.io

guardrail-dev /guardrail

Principled code generation from OpenAPI specifications

grarcht /Shuttle

Prevent TransactionTooLargeException crashes. For good. A modern Android framework for safely transporting large Serializable objects between components, without the crashes.

whitecircle /circle-guard-bench

First-of-its-kind AI benchmark for evaluating the protection capabilities of large language model (LLM) guard systems (guardrails and safeguards)

git-disl /Virus

This is the official code for the paper "Virus: Harmful Fine-tuning Attack for Large Language Models Bypassing Guardrail Moderation"

jomatsu /pi-jev-auto-mode

Jev (TypeSafe System One) backed auto mode for the Pi coding agent: semantically auto-approves bash, write, and edit tool calls and fails closed when a decision cannot be made.

infralicious /awesome-service-control-policies

Awesome AWS service control policies (SCPs), Resource Control Policies (RCPs), and other organizational policies

guardrail-dev /sbt-guardrail

Principled code generation from OpenAPI specifications

OMD-123 /agent-loop-guard

Zero-dependency guardrail framework for AI agents. Stops runaway loops, prompt injection, hallucination and unsafe tool calls before they execute. Provider-agnostic, dual ESM/CJS, TypeScript-first.

amirimi /Guardian

Extensible Go microkernel for LLM guardrails, quota enforcement and policy plugins.

phuryn /burnstop

A budget fuse for Claude Code: cap a session in tokens or dollars and halt runaway goals and loops. Auto-arms a $50 cap on goals.

skan0779 /korean-guardrail-dataset

Korean-Language Guardrail Evaluation Dataset for AI Agents

groundlens-dev /groundlens

Guardrails for LLM answers. GroundLens checks an LLM answer against its sources without a model, decides PASS, REVIEW or FAIL under your policy, and keeps a signed record.

Huzefaaa2 /terraform-guardrail

Terraform-Guardrail (TerraGuard) MCP is an open-source governance framework that enforces consistent, executable guardrails for Terraform across CI/CD pipelines—helping teams deliver secure, compliant infrastructure at scale without slowing down development.

chenflux /ClamAI

AI 安全护栏 — 智能体防护 · 内容安全 · 多模型代理 | 可部署为网关或终端 EDR

jake88nb-debug /hunt-guard

Outbound guardrail for authorized security testing. Blocks irreversible ops (delete/remove/drop/rm -rf) across curl, Python scripts, Burp, and AI agent sessions. Windows, zero deps, off unless you start it.

jiesou /dsh-stream-rules

模式匹配自动注入 steering rules，不占系统上下文 - Inject rules when needed, without wasting context. Similar to oh-my-pi's "Time-traveling stream rules", but with a very simple and compact code implementation.

kwangilkimkenny /dreamdojo-aegis-sdk

AEGIS DreamDojo Guard SDK — Runtime safety validation framework for embodied AI world models (NVIDIA DreamDojo / DiT pipelines)

guardrail-dev /guardrail-gradle-plugin

Principled code generation from OpenAPI specifications

nickfun /learning-guardrail

Example app of Akka-Http, Scala, Guardrail that implements the Todo Backend API