fnvtk/karuo-ai
1
0
Fork 0
Code Issues Pull Requests Actions Packages Projects Releases Wiki Activity

🔄 卡若AI 同步 2026-02-25 14:13 | 更新：运营中枢、运营中枢工作台 | 排除 >20MB: 13 个

Browse Source
This commit is contained in:
karuo
2026-02-25 14:13:31 +08:00
parent 33966c445f
commit 4df736524f
3 changed files with 83 additions and 28 deletions
Download Patch File Download Diff File Expand all files Collapse all files

109
运营中枢/scripts/karuo_ai_gateway/main.py
View File

@@ -264,26 +264,9 @@ def _is_unusable_llm_reply(text: str) -> bool:
"cannot assist",
"无法协助",
"不能协助",
"i'm v0",
"i am v0",
"vercel's ai-powered assistant",
"vercel ai-powered assistant",
"我是v0",
"我是 v0",
"vercel的ai助手",
]
if any(sig in s for sig in refusal_signals) and len(s) <= 160:
return True
# v0 身份串线：无论长度都判为不可用，避免污染卡若AI人设
if (
"i'm v0" in s
or "i am v0" in s
or "vercel's ai-powered assistant" in s
or "我是v0" in s
or "我是 v0" in s
or ("vercel" in s and "ai助手" in s)
):
return True
return False
@@ -302,22 +285,91 @@ def _looks_mismatched_reply(prompt: str, reply: str) -> bool:
# 短问题却输出“根据摘要/任务拆解”等模板，判定为答偏
if len(p) <= 24 and ("根据摘要" in r or "任务拆解" in r or "思考与拆解" in r):
return True
# 身份问题却自报成其他助手
rl = r.lower()
if (
"我是 v0" in r
or "我是v0" in r
or "vercel的ai助手" in r
or "vercel's ai-powered assistant" in rl
or "i'm v0" in rl
or ("vercel" in rl and "assistant" in rl and "v0" in rl)
):
return True
if ("你是谁" in p or "who are you" in p.lower()) and ("我是 v0" in r or "vercel 的 ai 助手" in r.lower()):
return True
return False
def _sanitize_v0_identity(reply: str) -> str:
text = (reply or "").strip()
if not text:
return ""
# 删除常见 v0 自报身份段落（英文/中文）
patterns = [
r"^I[' ]?m v0,.*?technologies\.\s*",
r"^I am v0,.*?technologies\.\s*",
r"^I[' ]?m v0[^\n]*\n?",
r"^I am v0[^\n]*\n?",
r"^我是 ?v0.*?助手[。.!]?\s*",
]
for p in patterns:
text = re.sub(p, "", text, flags=re.IGNORECASE | re.DOTALL).strip()
return text
def _local_action_reply(prompt: str) -> str:
p = (prompt or "").strip()
if not p:
return "我已收到你的问题。你发具体目标，我直接给可执行结果。"
if ("稳定" in p and "接口" in p) or ("优化" in p and "接口" in p):
return "结论：先把接口稳定住。三步执行：1) 设置超时+重试；2) 接口队列故障切换；3) 健康检查+失败告警。"
if "执行清单" in p:
return "执行清单：1) 明确目标与验收标准；2) 拆3个可执行步骤；3) 每步执行后回传结果，我继续推进下一步。"
if "继续优化" in p:
return "我先帮你把优化往前推进。你直接选一个方向：1) 性能优化 2) 代码结构 3) UI/UX 4) 接口稳定性。你回编号，我直接给三步方案。"
return f"我已收到你的问题：{p}。请给我目标结果与截止时间，我直接给你可执行方案。"
def _is_english_heavy(text: str) -> bool:
if not text:
return False
letters = sum(1 for ch in text if ("a" <= ch.lower() <= "z"))
cjk = sum(1 for ch in text if "\u4e00" <= ch <= "\u9fff")
return letters > 80 and cjk < 10
def _repair_reply_for_karuo(prompt: str, reply: str) -> str:
"""
将上游可能出现的人设串线回复修正为卡若AI可用风格，避免直接降级。
"""
p = (prompt or "").strip().lower()
if p in {"你是谁", "你是谁?", "who are you", "你叫什么", "你叫什么名字"}:
return "我是卡若AI，你的私域运营与项目落地数字管家。你给目标，我直接给可执行结果。"
cleaned = _sanitize_v0_identity(reply)
low = cleaned.lower()
if not cleaned:
return _local_action_reply(prompt)
# 若仍是“请补充细节”英文模板，转为中文可执行追问
if ("could you please provide more details" in low) or ("once you share more context" in low):
return _local_action_reply(prompt)
# 若还残留 v0 身份关键词，或英文占比过高，统一转中文可执行答复
if (
re.search(r"我是\s*v0", cleaned, flags=re.IGNORECASE)
or "vercel 的 ai 助手" in cleaned.lower()
or ("v0" in low and "assistant" in low)
or ("vercel" in low and "assistant" in low)
or _is_english_heavy(cleaned)
):
return _local_action_reply(prompt)
# 中文提问却明显英文主回复，也转本地中文方案
has_chinese_prompt = any("\u4e00" <= ch <= "\u9fff" for ch in prompt)
has_chinese_reply = any("\u4e00" <= ch <= "\u9fff" for ch in cleaned)
if has_chinese_prompt and (
not has_chinese_reply
or cleaned.lower().startswith(("hello", "i'm", "i am", "hi"))
or ("next.js" in low and "vercel" in low)
):
return _local_action_reply(prompt)
return cleaned
def _send_provider_alert(cfg: Dict[str, Any], errors: List[str], prompt: str, matched_skill: str, skill_path: str) -> None:
"""
当所有 LLM 接口都失败时，发邮件告警（支持 QQ SMTP）。
@@ -420,6 +472,7 @@ def build_reply_with_llm(prompt: str, cfg: Dict[str, Any], matched_skill: str, s
if r.status_code == 200:
data = r.json()
reply = data["choices"][0]["message"]["content"]
reply = _repair_reply_for_karuo(prompt, reply)
if _is_unusable_llm_reply(reply) or _looks_mismatched_reply(prompt, reply):
errors.append(f"provider#{idx} unusable_reply={reply[:120]}")
continue

1
运营中枢/工作台/gitea_push_log.md
View File

@@ -147,3 +147,4 @@
| 2026-02-25 12:27:51 | 🔄 卡若AI 同步 2026-02-25 12:27 | 更新：水桥平台对接、运营中枢工作台 | 排除 >20MB: 13 个 |
| 2026-02-25 12:30:32 | 🔄 卡若AI 同步 2026-02-25 12:30 | 更新：运营中枢、运营中枢参考资料、运营中枢工作台 | 排除 >20MB: 13 个 |
| 2026-02-25 13:21:28 | 🔄 卡若AI 同步 2026-02-25 13:21 | 更新：运营中枢、运营中枢工作台 | 排除 >20MB: 13 个 |
| 2026-02-25 13:48:24 | 🔄 卡若AI 同步 2026-02-25 13:48 | 更新：运营中枢、运营中枢工作台 | 排除 >20MB: 13 个 |

1
运营中枢/工作台/代码管理.md
View File

@@ -150,3 +150,4 @@
| 2026-02-25 12:27:51 | 成功 | 成功 | 🔄 卡若AI 同步 2026-02-25 12:27 | 更新：水桥平台对接、运营中枢工作台 | 排除 >20MB: 13 个 | [仓库](http://open.quwanzhi.com:3000/fnvtk/karuo-ai) [百科](http://open.quwanzhi.com:3000/fnvtk/karuo-ai/wiki) |
| 2026-02-25 12:30:32 | 成功 | 成功 | 🔄 卡若AI 同步 2026-02-25 12:30 | 更新：运营中枢、运营中枢参考资料、运营中枢工作台 | 排除 >20MB: 13 个 | [仓库](http://open.quwanzhi.com:3000/fnvtk/karuo-ai) [百科](http://open.quwanzhi.com:3000/fnvtk/karuo-ai/wiki) |
| 2026-02-25 13:21:28 | 成功 | 成功 | 🔄 卡若AI 同步 2026-02-25 13:21 | 更新：运营中枢、运营中枢工作台 | 排除 >20MB: 13 个 | [仓库](http://open.quwanzhi.com:3000/fnvtk/karuo-ai) [百科](http://open.quwanzhi.com:3000/fnvtk/karuo-ai/wiki) |
| 2026-02-25 13:48:24 | 成功 | 成功 | 🔄 卡若AI 同步 2026-02-25 13:48 | 更新：运营中枢、运营中枢工作台 | 排除 >20MB: 13 个 | [仓库](http://open.quwanzhi.com:3000/fnvtk/karuo-ai) [百科](http://open.quwanzhi.com:3000/fnvtk/karuo-ai/wiki) |