feat(safety): LLM healthy mode

[Soulter]

Modifications / 改动点

• This is NOT a breaking change. / 这不是一个破坏性变更。

Screenshots or Test Results / 运行截图或测试结果

---

Checklist / 检查清单

• 😊 如果 PR 中有新加入的功能，已经通过 Issue / 邮件等方式和作者讨论过。/ If there are new features added in the PR, I have discussed it with the authors through issues/emails, etc.
• 👀 我的更改经过了良好的测试，并已在上方提供了“验证步骤”和“运行截图”。/ My changes have been well-tested, and "Verification Steps" and "Screenshots" have been provided above.
• 🤓 我确保没有引入新依赖库，或者引入了新依赖库的同时将其添加到了 requirements.txt 和 pyproject.toml 文件相应位置。/ I have ensured that no new dependencies are introduced, OR if new dependencies are introduced, they have been added to the appropriate locations in requirements.txt and pyproject.toml.
• 😮 我的更改没有引入恶意代码。/ My changes do not introduce malicious code.

Summary by Sourcery

实现一个可选的 LLM 安全模式，在向服务提供方发送请求前预先添加一个以安全为重点的 system prompt，并在默认配置和 UI 元数据中暴露其配置选项。

New Features:

• 在默认设置和服务提供方配置元数据中新增可配置的 LLM 安全模式和策略选项。
• 引入一个安全模式的 system prompt，引导模型给出安全、健康的回复，同时拒绝有害或敏感的请求。

Enhancements:

• 将 LLM 安全模式集成到内部代理的处理流水线中，使启用该模式的提供方在处理请求前自动接收安全 system prompt。

Original summary in English

Summary by Sourcery

Implement an optional LLM safety mode that prepends a safety-focused system prompt to provider requests and expose its configuration in defaults and UI metadata.

New Features:

• Add configurable LLM safety mode and strategy options to the default settings and provider configuration metadata.
• Introduce a safety mode system prompt that guides models toward safe, healthy responses while refusing harmful or sensitive requests.

Enhancements:

• Integrate LLM safety mode into the internal agent processing pipeline so that enabled providers automatically receive the safety system prompt before handling requests.

[sourcery-ai]

Hey - 我在这里给出了一些整体层面的反馈：

• 在 _apply_llm_safety_mode 中，建议考虑是否应当在已有的 req.system_prompt 之后 追加安全提示（或者在文档中明确说明安全提示会优先生效），以避免意外改变现有 system 指令的相对权重/优先级。
• 当 safety_mode_strategy 无法识别时，你目前只记录了一条警告日志，并且实际上对该请求禁用了安全模式；建议在这种情况下回退到默认的 system_prompt 策略，或者快速失败（fail fast），以避免因为配置错误而在悄无声息中绕过安全机制。

给 AI 代理的提示词

Please address the comments from this code review:

## Overall Comments
- In `_apply_llm_safety_mode`, consider whether the safety prompt should be appended after any existing `req.system_prompt` (or clearly documented as taking top priority) to avoid unexpectedly changing the relative weight/priority of existing system instructions.
- When `safety_mode_strategy` is not recognized, you currently only log a warning and effectively disable safety mode for that request; consider falling back to the default `system_prompt` strategy or failing fast so misconfiguration does not silently bypass the safety behavior.

---

Sourcery 对开源项目免费 —— 如果你觉得我们的评审有帮助，欢迎分享给更多人 ✨

• X
• Mastodon
• LinkedIn
• Facebook

_{帮我变得更有用！请在每条评论上点击 👍 或 👎，我会根据你的反馈改进后续的评审质量。}

Original comment in English

Hey - I've left some high level feedback:

• In _apply_llm_safety_mode, consider whether the safety prompt should be appended after any existing req.system_prompt (or clearly documented as taking top priority) to avoid unexpectedly changing the relative weight/priority of existing system instructions.
• When safety_mode_strategy is not recognized, you currently only log a warning and effectively disable safety mode for that request; consider falling back to the default system_prompt strategy or failing fast so misconfiguration does not silently bypass the safety behavior.

Prompt for AI Agents

Please address the comments from this code review:

## Overall Comments
- In `_apply_llm_safety_mode`, consider whether the safety prompt should be appended after any existing `req.system_prompt` (or clearly documented as taking top priority) to avoid unexpectedly changing the relative weight/priority of existing system instructions.
- When `safety_mode_strategy` is not recognized, you currently only log a warning and effectively disable safety mode for that request; consider falling back to the default `system_prompt` strategy or failing fast so misconfiguration does not silently bypass the safety behavior.

---

Sourcery is free for open source - if you like our reviews please consider sharing them ✨

• X
• Mastodon
• LinkedIn
• Facebook

_{Help me be more useful! Please click 👍 or 👎 on each comment and I'll use the feedback to improve your reviews.}