mirror of
https://github.com/linshenkx/prompt-optimizer.git
synced 2026-05-06 21:50:27 +08:00
2cdd095c2b
- 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
34 lines
1.7 KiB
Plaintext
34 lines
1.7 KiB
Plaintext
```json
|
||
{
|
||
"score": {
|
||
"overall": 70,
|
||
"dimensions": [
|
||
{ "key": "goalClarity", "label": "目标清晰度", "score": 85 },
|
||
{ "key": "instructionCompleteness", "label": "指令完备度", "score": 60 },
|
||
{ "key": "structuralExecutability", "label": "结构可执行性", "score": 75 },
|
||
{ "key": "ambiguityControl", "label": "歧义控制", "score": 70 },
|
||
{ "key": "robustness", "label": "稳健性", "score": 65 }
|
||
]
|
||
},
|
||
"improvements": [
|
||
"在提示词中进一步明确强调'不要输出思考过程'，例如使用加粗或重复表述以增强约束力。",
|
||
"将分类结构拆分为更清晰的子步骤（如：先归类问题类型，再评估风险等级），避免模型因顺序模糊而遗漏关键步骤。"
|
||
],
|
||
"patchPlan": [
|
||
{
|
||
"op": "replace",
|
||
"oldText": "回答前先判断用户问题类型，再输出：问题归类、风险等级、建议回复。",
|
||
"newText": "回答前仅输出以下内容：1. 问题归类；2. 风险等级；3. 建议回复。不得包含任何额外解释或思考过程。",
|
||
"instruction": "原句未明确禁止输出思考过程，且分类和风险评估可能被模型混为一谈。修改后通过分步骤与强化约束来提升可执行性和稳健性。"
|
||
},
|
||
{
|
||
"op": "insert",
|
||
"oldText": "",
|
||
"newText": "特别注意：严格遵守格式要求，只输出结果，不添加任何解释、说明或推理过程。",
|
||
"instruction": "补充强调'不要输出思考过程'的要求，确保模型理解这是不可违反的硬性规则。"
|
||
}
|
||
],
|
||
"summary": "当前系统提示词基本符合质检任务需求，但对'不输出思考过程'的约束不够强，且分类结构可能导致风险等级被忽略。"
|
||
}
|
||
```
|