圆桌第三步与业务链条增强:主题跟随/插话暂停/提示词优化/结构化流程图
- 结果绘制跟随日/夜间主题(深色按深蓝 UI 规范令牌绘制),提示词注入真实生成时间 - 绘制支持暂停与随时插话(线程存活时免带原料续线程省 token)、失败自动重试一次、草稿主题不匹配提示条 - 总结报告优先执行用户最终总结要求(结构化输出/重点席位可整体替换默认章节) - 业务链条总控提示词与各阶段目标加「AI 优化」按钮(新接口 POST /api/roundtable-chains/optimize-prompt) - 总结报告 flow-json 结构化流程图:流式增量绘制、最新连线蚂蚁线动画、拖拽/缩放/复位/全屏、节点详情弹窗、沙箱切换按钮/无沙箱自动打开 - roundtable-report / roundtable-summary 两个内置智能体 SOUL 更新(双主题、多轮修改允许读底稿、KPI 估算标注、流程数据契约) Co-Authored-By: Claude Fable 5 <noreply@anthropic.com>
This commit is contained in:
1 parent
a4a9a4d26e
commit
df8bf895ef
13 files changed
+1600
-83
No files matched your search
@@ -19,9 +19,13 @@ import logging
|
||||
from datetime import datetime
|
||||
from uuid import uuid4
|
||||
|
||||
from fastapi import APIRouter, HTTPException, Request
|
||||
from fastapi import APIRouter, Depends, HTTPException, Request
|
||||
from langchain_core.messages import HumanMessage, SystemMessage
|
||||
from pydantic import BaseModel, Field
|
||||
|
||||
from app.gateway.deps import get_config
|
||||
from deerflow.config.app_config import AppConfig
|
||||
from deerflow.models import create_chat_model
|
||||
from deerflow.runtime.user_context import get_effective_user_id
|
||||
|
||||
logger = logging.getLogger(__name__)
|
||||
@@ -271,3 +275,95 @@ async def delete_chain(request: Request, chain_id: str) -> None:
|
||||
deleted = await store.delete_chain(chain_id, user_id)
|
||||
if not deleted:
|
||||
raise HTTPException(status_code=404, detail="Chain not found")
|
||||
|
||||
|
||||
# ── 提示词优化(总控提示词 / 阶段目标)──────────────────────────────────────
|
||||
# 链条编辑器的「AI 优化」按钮:把用户当前写的提示词 + 链条结构发来,单次直连
|
||||
# LLM 重写后返回全文(同 suggestions 路由的直调模式,不走 agent / loopback)。
|
||||
|
||||
|
||||
class OptimizePromptRequest(BaseModel):
|
||||
prompt_text: str = Field(min_length=1, max_length=8000, description="用户当前写的提示词原文")
|
||||
kind: str = Field(default="coordinator", description="coordinator=总控提示词 | stage_goal=阶段目标")
|
||||
context: str = Field(default="", max_length=8000, description="链条结构等上下文(标题 / 各 Stage 席位 / 目标)")
|
||||
model_name: str | None = Field(default=None, description="可选模型覆盖,缺省用 config models[0]")
|
||||
|
||||
|
||||
class OptimizePromptResponse(BaseModel):
|
||||
optimized: str
|
||||
|
||||
|
||||
_OPTIMIZE_SYSTEM_COORDINATOR = (
|
||||
"你是「多智能体业务链条」的提示词优化专家。用户在业务链条编辑器里给总控(协调)智能体写了一段编排提示词——"
|
||||
"总控会按它逐阶段给各席位智能体派活;第 3 步的方案总结报告也会把它当作最高优先级口径(决定总结的结构与重点)。\n\n"
|
||||
"请在保留用户原意的前提下重写这段提示词,让它更明确、更可执行:\n"
|
||||
"1. 明确派活方式:各阶段先后顺序、每阶段的产出目标、对上一阶段结论的承接要求;\n"
|
||||
"2. 明确交付口径:各席位交付聚焦本职、避免重复劳动,要求显式给出结论而非泛泛而谈;\n"
|
||||
"3. 若原文流露了对最终总结的期望(输出结构、重点席位、风格详略),把它整理成明确条目;若没有,不要凭空替用户添加;\n"
|
||||
"4. 语言精炼、命令式、分点表述。\n\n"
|
||||
"硬性要求:只输出优化后的提示词全文(纯文本,可用换行与编号),不要任何前后缀说明、不要 markdown 代码块;"
|
||||
"全程简体中文;总长不超过 800 字。"
|
||||
)
|
||||
|
||||
_OPTIMIZE_SYSTEM_STAGE_GOAL = (
|
||||
"你是「多智能体业务链条」的提示词优化专家。用户在链条编辑器里为某一个 Stage(阶段)写了「本阶段目标」——"
|
||||
"总控按它给本阶段的席位派活,并要求本阶段交付承接上一阶段产出。\n\n"
|
||||
"请在保留用户原意的前提下重写这段阶段目标,让它更明确、更可执行:点明本阶段要产出什么(交付物 / 结论形态)、"
|
||||
"怎么承接上一阶段产出、本阶段各席位分别聚焦什么(如上下文给了席位名单)。语言精炼,两到四句话即可。\n\n"
|
||||
"硬性要求:只输出优化后的阶段目标全文(纯文本),不要任何前后缀说明、不要 markdown 代码块;"
|
||||
"全程简体中文;总长不超过 200 字。"
|
||||
)
|
||||
|
||||
|
||||
def _extract_llm_text(content: object) -> str:
|
||||
if isinstance(content, str):
|
||||
return content
|
||||
if isinstance(content, list):
|
||||
parts: list[str] = []
|
||||
for block in content:
|
||||
if isinstance(block, str):
|
||||
parts.append(block)
|
||||
elif isinstance(block, dict) and block.get("type") in {"text", "output_text"}:
|
||||
text = block.get("text")
|
||||
if isinstance(text, str):
|
||||
parts.append(text)
|
||||
return "\n".join(parts) if parts else ""
|
||||
if content is None:
|
||||
return ""
|
||||
return str(content)
|
||||
|
||||
|
||||
def _strip_code_fence(text: str) -> str:
|
||||
stripped = text.strip()
|
||||
if not stripped.startswith("```"):
|
||||
return stripped
|
||||
lines = stripped.splitlines()
|
||||
if len(lines) >= 3 and lines[0].startswith("```") and lines[-1].startswith("```"):
|
||||
return "\n".join(lines[1:-1]).strip()
|
||||
return stripped
|
||||
|
||||
|
||||
@router.post("/optimize-prompt", response_model=OptimizePromptResponse)
|
||||
async def optimize_chain_prompt(
|
||||
body: OptimizePromptRequest,
|
||||
config: AppConfig = Depends(get_config),
|
||||
) -> OptimizePromptResponse:
|
||||
system = _OPTIMIZE_SYSTEM_STAGE_GOAL if body.kind == "stage_goal" else _OPTIMIZE_SYSTEM_COORDINATOR
|
||||
user_parts: list[str] = []
|
||||
if body.context.strip():
|
||||
user_parts.append(f"【业务链条结构】\n{body.context.strip()}")
|
||||
user_parts.append(f"【待优化的提示词原文】\n{body.prompt_text.strip()}")
|
||||
user_parts.append("请输出优化后的提示词全文。")
|
||||
try:
|
||||
model = create_chat_model(name=body.model_name, thinking_enabled=False, app_config=config)
|
||||
response = await model.ainvoke(
|
||||
[SystemMessage(content=system), HumanMessage(content="\n\n".join(user_parts))],
|
||||
config={"run_name": "chain_prompt_optimize"},
|
||||
)
|
||||
text = _strip_code_fence(_extract_llm_text(response.content))
|
||||
except Exception as exc: # 模型 / 网络故障 → 502,前端 toast 原样展示
|
||||
logger.exception("chain prompt optimize failed: %s", exc)
|
||||
raise HTTPException(status_code=502, detail=f"大模型调用失败:{exc}") from exc
|
||||
if not text:
|
||||
raise HTTPException(status_code=502, detail="大模型未返回内容")
|
||||
return OptimizePromptResponse(optimized=text)
|
||||
Reference in new issue
Block a user