Microsoft archived PyRIT (Mar 2026) - what LLM red-teamers should use instead

微软已归档 PyRIT（2026年3月）—— LLM 红队测试人员该用什么替代？

Quick one: if PyRIT (Microsoft’s Python Risk Identification Tool) is on your shortlist for LLM red-teaming, check the repo first. Azure/PyRIT was archived on GitHub on March 27, 2026. It’s read-only now: no commits, no releases, no issue triage, nothing. Whatever version you pip-installed is the last version you’ll ever get. 简要说明：如果 PyRIT（微软的 Python 风险识别工具）在你的 LLM 红队测试工具清单中，请先检查一下代码库。Azure/PyRIT 已于 2026 年 3 月 27 日在 GitHub 上归档。它现在处于只读状态：没有提交、没有发布、没有问题分类，什么都没有。你通过 pip 安装的任何版本都将是你最后能用到的版本。

That matters more for PyRIT than it would for most tools, because PyRIT was never a turnkey scanner. It’s a framework for scripting multi-turn attack orchestration, the kind of thing a red team builds custom attack sequences on top of. A framework that’s stopped shipping fixes is a worse foundation to build on than a finished tool that’s stopped shipping features, because you were relying on it staying flexible to your needs, and now it can’t. 这对 PyRIT 的影响比大多数工具都要大，因为 PyRIT 从来都不是一个“开箱即用”的扫描器。它是一个用于编写多轮攻击编排脚本的框架，红队通常在其基础上构建自定义的攻击序列。一个停止发布修复程序的框架，作为构建基础时，比一个停止发布新功能的成品工具更糟糕，因为你曾依赖它保持灵活性以满足你的需求，而现在它做不到了。

So what do you use instead? Depends on what you were actually using PyRIT for: 那么，你应该用什么来替代呢？这取决于你使用 PyRIT 的具体目的：

You wanted a broad, actively maintained app-layer scanner -> promptfoo. Zero-install via npx promptfoo, 50+ red-team plugins, OWASP/NIST/MITRE ATLAS report mappings, and it’s still getting regular releases. 如果你需要一个广泛且持续维护的应用层扫描器 -> promptfoo。通过 npx promptfoo 零安装，拥有 50 多个红队插件，支持 OWASP/NIST/MITRE ATLAS 报告映射，并且仍在定期发布更新。

You wanted model-layer testing (jailbreaks, encoding tricks, data leakage on the base model itself, not your app) -> garak. NVIDIA-maintained, pip installable, 8k+ stars, actively developed. 如果你需要模型层测试（越狱、编码技巧、基础模型本身的数据泄露，而非针对你的应用） -> garak。由 NVIDIA 维护，可通过 pip 安装，拥有 8k+ 星标，且处于活跃开发状态。

You wanted OWASP-mapped detectors and don’t mind a paid tier for continuous scanning -> Giskard. The open source scanner is real and current; the always-on Hub is commercial. 如果你需要基于 OWASP 映射的检测器，并且不介意为持续扫描付费 -> Giskard。其开源扫描器真实且保持更新；而“始终在线”的 Hub 是商业服务。

You wanted a fast, zero-setup smoke test before reaching for any of the above -> that’s the gap we built sentinel-scan-cli for. Dependency-free CLI (Python and npm ports, identical output), 15 attack patterns each tagged to its OWASP LLM Top 10 category, —demo runs with no config and no API keys in under a minute. 如果你在尝试上述工具前，需要一个快速、零配置的冒烟测试 -> 这正是我们构建 sentinel-scan-cli 的初衷。无依赖的 CLI（提供 Python 和 npm 版本，输出一致），包含 15 种攻击模式，每种都标记了对应的 OWASP LLM Top 10 类别，使用 --demo 参数可在不到一分钟内运行，无需配置，无需 API 密钥。

None of these replace PyRIT’s specific multi-turn orchestration model one-for-one, if that’s genuinely what you need, Microsoft’s PyRIT Community fork discussion or building your own harness on top of a maintained model API is probably the honest answer now. But if you were reaching for PyRIT as a general “test my LLM thing for prompt injection” tool, one of the four above almost certainly covers your case and is still getting updates. 这些工具都无法完全一对一地替代 PyRIT 特有的多轮编排模型。如果你确实需要这种功能，微软的 PyRIT 社区分支讨论，或者在受维护的模型 API 之上构建自己的测试工具，可能是目前最诚实的答案。但如果你只是把 PyRIT 当作一个通用的“测试我的 LLM 是否存在提示词注入”的工具，那么上述四个工具几乎肯定能满足你的需求，且它们仍在持续更新。

We went deeper on all five (well, four now) tools, what layer they test, and where each one actually fits, in the full comparison guide, updated today to reflect PyRIT’s archived status. Full disclosure: I work on sentinel-scan-cli, so weigh that mention accordingly. 我们在完整的对比指南中深入探讨了所有五个（现在是四个）工具，包括它们测试的层级以及各自的适用场景。该指南已于今日更新，以反映 PyRIT 的归档状态。披露：我参与了 sentinel-scan-cli 的开发，请在参考该建议时予以考量。