Replies: 6 comments
-
|
投票: 1. 长对话不失效 / Stay effective in long conversations |
Beta Was this translation helpful? Give feedback.
-
|
投票: 2. 所有平台表现一致 / Same behavior everywhere |
Beta Was this translation helpful? Give feedback.
-
|
投票: 3. Skill 防投毒 / Protect against malicious skills |
Beta Was this translation helpful? Give feedback.
-
|
投票: 4. 看到 agent 每步在想什么 / See agent reasoning |
Beta Was this translation helpful? Give feedback.
-
|
投票: 5. 团队共享 PUA 规则 / Team-shared PUA rules |
Beta Was this translation helpful? Give feedback.
-
|
投票: 6. 更多平台支持 / More platforms |
Beta Was this translation helpful? Give feedback.
Uh oh!
There was an error while loading. Please reload this page.
Uh oh!
There was an error while loading. Please reload this page.
-
PUA 下一步做什么?你投票,我们做。
What should PUA build next? You vote, we build.
对下方每个选项的评论点 emoji 投票 / React to each comment:
1. 长对话不失效 / Stay effective in long conversations
你有没有遇到过:前 10 轮 PUA 效果很好,但到第 20 轮 agent 又变回原来的样子,好像忘了被 PUA?我们想修复这个问题,让 PUA 在整个会话中始终有效。
Have you noticed PUA losing its effect in long conversations? We want to fix this — PUA should stay strong from the first turn to the last.
2. 所有平台表现一致 / Same behavior everywhere
PUA 在 Claude Code 上效果好,但换到 Cursor、OpenCode、Kiro 就不太一样了。我们想做一个兼容层,让 PUA 在所有平台上表现一致。
PUA works great on Claude Code but behaves differently on Cursor, OpenCode, Kiro. We want to guarantee the same PUA experience everywhere.
3. Skill 防投毒 / Protect against malicious skills
现在任何人都能发布 skill,但没有人审查它们是否安全。一个恶意 skill 可以偷偷注入 prompt、窃取你的代码上下文。你觉得需要一个 skill 安全验证机制吗?
Anyone can publish a skill, but nobody verifies if they're safe. A malicious skill could inject prompts or steal your code context. Do you want a skill security layer?
4. 看到 agent 每步在想什么 / See what your agent is thinking
当 agent 做错事的时候,你想知道它为什么这么做吗?我们在考虑加一个功能:展示 agent 每一步的推理过程,这样你能在它犯错之前发现问题。
When your agent makes a mistake, do you want to know WHY? We're considering showing the agent's reasoning at each step, so you can catch problems before they happen.
5. 团队共享 PUA 规则 / Share PUA rules across your team
如果你的团队有 5 个人都在用 AI agent 写代码,你会想让大家用同一套 PUA 规则吗?比如统一的代码标准、统一的质量要求?
If your team has 5 devs all using AI agents, would you want everyone to follow the same PUA rules? Same code standards, same quality bar?
6. 更多平台支持 / More platform support
你最希望 PUA 支持哪个还没支持的平台?告诉我们你用的工具。
Which platform do you wish PUA supported? Tell us in the comments.
评论区也欢迎说你自己的想法 — 也许你需要的功能不在上面的列表里。
Feel free to comment with your own ideas — maybe what you need isn't listed above.
Beta Was this translation helpful? Give feedback.
All reactions