mirror of
https://github.com/linshenkx/prompt-optimizer.git
synced 2026-07-01 23:44:42 +08:00
- fold earlier planning notes into a single current-spec and archived history structure - keep manual acceptance steps and real API samples aligned with the refactored analysis/result/compare model - retain supporting workspace notes needed to review version-selection and evaluation behavior changes
2.4 KiB
2.4 KiB
basic-user-prompt-only
Meta
- Type:
prompt-only - Model:
dashscope - Duration:
8653ms - Timestamp:
1773729077366
Score
- Overall:
85
| Key | Label | Score |
|---|---|---|
goalClarity |
目标清晰度 | 90 |
instructionCompleteness |
指令完备度 | 85 |
structuralExecutability |
结构可执行性 | 90 |
ambiguityControl |
歧义控制 | 80 |
robustness |
稳健性 | 80 |
Improvements
- 将'不要解释'改为更强烈的表述,例如'禁止添加任何说明性文字或尾注',以明确约束输出格式。
Patch Plan
Patch 1
- Op:
replace - Instruction: 当前'不要解释'可能不足以阻止模型在输出后追加解释性内容。强化该指令为'禁止添加任何说明性文字或尾注',以确保模型严格遵守输出要求。
Old Text
不要解释。
New Text
禁止添加任何说明性文字或尾注。
Summary
提示词整体设计良好,但'不要解释'的表达力度较弱,可能诱发模型附加说明性内容。
Raw Response JSON
{
"type": "prompt-only",
"score": {
"overall": 85,
"dimensions": [
{
"key": "goalClarity",
"label": "目标清晰度",
"score": 90
},
{
"key": "instructionCompleteness",
"label": "指令完备度",
"score": 85
},
{
"key": "structuralExecutability",
"label": "结构可执行性",
"score": 90
},
{
"key": "ambiguityControl",
"label": "歧义控制",
"score": 80
},
{
"key": "robustness",
"label": "稳健性",
"score": 80
}
]
},
"improvements": [
"将'不要解释'改为更强烈的表述,例如'禁止添加任何说明性文字或尾注',以明确约束输出格式。"
],
"summary": "提示词整体设计良好,但'不要解释'的表达力度较弱,可能诱发模型附加说明性内容。",
"patchPlan": [
{
"op": "replace",
"oldText": "不要解释。",
"newText": "禁止添加任何说明性文字或尾注。",
"instruction": "当前'不要解释'可能不足以阻止模型在输出后追加解释性内容。强化该指令为'禁止添加任何说明性文字或尾注',以确保模型严格遵守输出要求。"
}
],
"metadata": {
"model": "dashscope",
"timestamp": 1773729077366,
"duration": 8653
}
}