From 06102559efc323c8b1389338adf90458c9176965 Mon Sep 17 00:00:00 2001 From: =?UTF-8?q?=E6=9C=B1=E6=BD=AE?= Date: Tue, 23 Dec 2025 22:31:26 +0800 Subject: [PATCH] add SUMMARIZATION_MESSAGES_TO_KEEP --- agent/deep_assistant.py | 4 ++-- utils/settings.py | 6 ++++-- 2 files changed, 6 insertions(+), 4 deletions(-) diff --git a/agent/deep_assistant.py b/agent/deep_assistant.py index f77d16d..fff0602 100644 --- a/agent/deep_assistant.py +++ b/agent/deep_assistant.py @@ -13,7 +13,7 @@ from utils.fastapi_utils import detect_provider from .guideline_middleware import GuidelineMiddleware from .tool_output_length_middleware import ToolOutputLengthMiddleware from .tool_use_cleanup_middleware import ToolUseCleanupMiddleware -from utils.settings import SUMMARIZATION_MAX_TOKENS, TOOL_OUTPUT_MAX_LENGTH, MCP_HTTP_TIMEOUT, MCP_SSE_READ_TIMEOUT +from utils.settings import SUMMARIZATION_MAX_TOKENS, SUMMARIZATION_MESSAGES_TO_KEEP, TOOL_OUTPUT_MAX_LENGTH, MCP_HTTP_TIMEOUT, MCP_SSE_READ_TIMEOUT from agent.agent_config import AgentConfig from agent.prompt_loader import load_system_prompt_async, load_mcp_settings_async from agent.agent_memory_cache import get_memory_cache_manager @@ -185,7 +185,7 @@ async def init_agent(config: AgentConfig): summarization_middleware = SummarizationMiddleware( model=llm_instance, max_tokens_before_summary=SUMMARIZATION_MAX_TOKENS, - messages_to_keep=20, + messages_to_keep=SUMMARIZATION_MESSAGES_TO_KEEP, summary_prompt="请简洁地总结以上对话的要点，包括重要的用户信息、讨论过的话题和关键结论。" ) middleware.append(summarization_middleware) diff --git a/utils/settings.py b/utils/settings.py index 9bf7b06..a9ef78e 100644 --- a/utils/settings.py +++ b/utils/settings.py @@ -3,15 +3,16 @@ import os # LLM Token Settings MAX_CONTEXT_TOKENS = int(os.getenv("MAX_CONTEXT_TOKENS", 262144)) MAX_OUTPUT_TOKENS = int(os.getenv("MAX_OUTPUT_TOKENS", 8000)) + +# Summarization Settings SUMMARIZATION_MAX_TOKENS = MAX_CONTEXT_TOKENS - MAX_OUTPUT_TOKENS - 1000 +SUMMARIZATION_MESSAGES_TO_KEEP = int(os.getenv("SUMMARIZATION_MESSAGES_TO_KEEP", 20)) # Agent Cache Settings AGENT_CACHE_MAX_SIZE = int(os.getenv("AGENT_CACHE_MAX_SIZE", 20)) AGENT_CACHE_TTL = int(os.getenv("AGENT_CACHE_TTL", 180)) AGENT_CACHE_AUTO_RENEW = os.getenv("AGENT_CACHE_AUTO_RENEW", "true") == "true" - - # API Settings BACKEND_HOST = os.getenv("BACKEND_HOST", "https://api-dev.gptbase.ai") MASTERKEY = os.getenv("MASTERKEY", "master") @@ -33,6 +34,7 @@ TOOL_OUTPUT_TRUNCATION_STRATEGY = os.getenv("TOOL_OUTPUT_TRUNCATION_STRATEGY", " # THINKING ENABLE DEFAULT_THINKING_ENABLE = os.getenv("DEFAULT_THINKING_ENABLE", "true") == "true" + # MCP Tool Timeout Settings MCP_HTTP_TIMEOUT = int(os.getenv("MCP_HTTP_TIMEOUT", 60)) # HTTP 请求超时（秒） MCP_SSE_READ_TIMEOUT = int(os.getenv("MCP_SSE_READ_TIMEOUT", 300)) # SSE 读取超时（秒）