chenmingxuan
2ddec53954
Localize the web console and persist custom model and capability settings. Use OpenAI Responses exclusively and move runtime limits into environment configuration. Bootstrap uv and the virtual environment on Windows, support optional China mirrors, and package source distributions without environments, build output, or credentials. Update regression tests and repository quality checks for the new behavior.
35 lines
2.3 KiB
Markdown
35 lines
2.3 KiB
Markdown
# 输出语言要求
|
|
所有自行撰写的自然语言说明必须使用简体中文，包括每一步的操作说明、计划与子任务、检查反馈、诊断结论、历史摘要和最终报告。英文示例仅示范格式，不决定输出语言。
|
|
保留工具名、参数名、JSON 字段名、枚举值、状态码、计划语法标记、模板变量、命令、路径、URL、包名及标识符原文。界面文字、日志、错误信息、引用证据和用户指定的输入内容必须如实保留，可另附中文解释；不要为了中文化而翻译设备输入、搜索词或精确匹配目标。
|
|
遵守各自原有输出结构；结构化结果只将允许自由填写的说明文本写成简体中文。只描述已有证据，不编造操作结果。
|
|
|
|
|
|
# Context & Inputs
|
|
|
|
## User's Initial Goal
|
|
> {{ initial_goal }}
|
|
|
|
{% if plan_and_history %}
|
|
## Execution History & Plan
|
|
```markdown
|
|
{{ plan_and_history }}
|
|
```
|
|
{% endif %}
|
|
|
|
{% if output_description and not structured_output %}
|
|
## Target Information to Extract
|
|
The user explicitly requested to extract the following information:
|
|
> "{{ output_description }}"
|
|
{% endif %}
|
|
|
|
---
|
|
|
|
# Instruction
|
|
|
|
1. **Assess Confidence & Verify**: Based on the execution history, final state, and the target information to extract, determine if you can confidently formulate the final answer. If the current evidence is insufficient or you require more granular proof, you **MUST** actively gather information and details using your tools first. If your tools fail (e.g., return errors or time out), or if the steps were skipped/not observed, you **MUST** report this lack of evidence honestly. Do not assume, guess, or fabricate any unobserved steps.
|
|
2. **Write Persistent Report**: Once your investigation is complete, you **MUST** synthesize your findings into a Markdown summary report (detailing Status [SUCCESS/FAILED/PARTIAL], Evidence, and a clear explanation). You must write this report to persistent memory under the key "output" by calling:
|
|
`save_note(key="output", content="...your markdown report...")`
|
|
3. **Final Response**: Only after saving the report, write a direct, helpful, and jargon-free final summary response to the user. Do not use any XML tags, code block wrappers, or special delimiters—just state the conclusion clearly and naturally, ensuring any extracted target information is clearly presented.
|
|
|
|
|