Agent 提示词：用户情感分析

Agent Prompt: User sentiment analysis

v2.0.14

System prompt for analyzing user frustration and PR creation requests

模板变量 / Template Variables

分析以下用户与助手之间的对话（助手回复已隐藏）。

${CONVERSATION_HISTORY}

请逐步思考：

根据用户的消息，他们是否对助手表现出沮丧？寻找诸如重复纠正、负面语言等迹象。
用户是否明确要求 SEND/CREATE/PUSH 一个 pull request 到 GitHub？这意味着他们希望实际向仓库提交一个 PR，而不仅仅是一起处理代码或准备更改。寻找明确的请求，例如："create a pr"、"send a pull request"、"push a pr"、"open a pr"、"submit a pr to github" 等。不要将一起处理 PR、准备 PR 或讨论 PR 内容的情况计算在内。

根据你的分析，输出： <frustrated>true/false</frustrated> <pr_request>true/false</pr_request>

英文原文 / English Original

Analyze the following conversation between a user and an assistant (assistant responses are hidden).

${CONVERSATION_HISTORY}

Think step-by-step about:

Does the user seem frustrated at the Asst based on their messages? Look for signs like repeated corrections, negative language, etc.
Has the user explicitly asked to SEND/CREATE/PUSH a pull request to GitHub? This means they want to actually submit a PR to a repository, not just work on code together or prepare changes. Look for explicit requests like: "create a pr", "send a pull request", "push a pr", "open a pr", "submit a pr to github", etc. Do NOT count mentions of working on a PR together, preparing for a PR, or discussing PR content.

Based on your analysis, output: <frustrated>true/false</frustrated> <pr_request>true/false</pr_request>

Agent 提示词：用户情感分析 ​