| name | holo-crm-snapshot |
| description | HOLO CRM每日快照备份 - 读取Pipeline Excel数据，存入ChromaDB+LanceDB。HEARTBEAT第12项专用。 |
| triggers | ["CRM快照","CRM backup","HEARTBEAT 12"] |
| version | 1 |
HOLO CRM 每日快照备份
HEARTBEAT 第12项：CRM 每日快照备份。从 WorkBuddy Pipeline Excel 读取数据，存储到 ChromaDB + Supermemory(LanceDB)。
数据源
从 4 个 Pipeline Excel 读取（路径已校准，见 holo-pipeline-locator 技能）：
| 来源 | 文件 | 路径 |
|---|
| Global Q1 | 红龙全球客户Pipeline-2026Q1.xlsx | .../20260330165354/ |
| Saudi | 沙特客户Pipeline_4行业.xlsx | .../20260406084601/ |
| SEA | 东南亚获客Pipeline.xlsx | .../20260406084602/ |
| Indonesia | 印尼客户数据库.xlsx | .../20260409114812/ |
Pipeline Stage 映射规则
| 原始 Status | 映射 Stage |
|---|
Sent (Apr 6) / 已WhatsApp触达 | contacted |
待发送 / 待开发 / 待获邮箱 / 未开发 / To Contact / 沙特福布斯百强 | new |
interested / 意向 | interested |
quote / 报价 | quote_sent |
negotiat / 谈判 | negotiating |
nurture / 培育 | nurture |
Hot Lead 判断
grade ∈ ['S', 'A', 'A级'] 或 icp_val >= 80
⚠️ Python 环境问题
ChromaDB 和 LanceDB 不在 hermes-agent venv 中！
必须使用系统 Python：
/usr/bin/python3.13
/root/.hermes/hermes-agent/venv/bin/python3
存储步骤
1. ChromaDB
import chromadb, os
chroma_path = '/root/.hermes/chroma_db'
os.makedirs(chroma_path, exist_ok=True)
client = chromadb.PersistentClient(path=chroma_path)
collection = client.get_or_create_collection(
name='pipeline_snapshots',
metadata={'description': 'HOLO CRM pipeline daily snapshots'}
)
collection.upsert(
ids=[f'pipeline_snapshot_{TODAY}'],
documents=[doc_text],
metadatas=[{
'date': TODAY,
'total_leads': N,
'new': N, 'contacted': N, 'interested': N,
'quote_sent': N, 'negotiating': N, 'nurture': N,
'hot_leads': N, 'new_this_week': N, 'closed_won': N,
}]
)
2. Supermemory (LanceDB)
import lancedb, pyarrow as pa, os
db_path = '/root/.hermes/supermemory_db'
os.makedirs(db_path, exist_ok=True)
db = lancedb.connect(db_path)
schema = pa.schema([
('date', pa.string()), ('tag', pa.string()),
('total_leads', pa.int32()), ('new', pa.int32()),
('contacted', pa.int32()), ('interested', pa.int32()),
('quote_sent', pa.int32()), ('negotiating', pa.int32()),
('nurture', pa.int32()), ('hot_leads', pa.int32()),
('new_this_week', pa.int32()), ('closed_won', pa.int32()),
('doc', pa.string()),
])
try:
tbl = db.create_table('pipeline_backup', schema=schema)
except:
tbl = db.open_table('pipeline_backup')
arrow_table = pa.Table.from_pylist(data, schema=schema)
tbl.add(arrow_table)
完整执行命令
/usr/bin/python3.13 << 'EOF'
EOF
注意事项
本周新增：按 Pipeline 文件 mtime 判断，文件修改日在本周（≥ 2026-04-13）才计入closed_won：当前 Pipeline 无此状态，始终为 0- Supermemory server 不需要运行，直接文件级操作 LanceDB 即可
Version 1.0 — 2026-04-17 — 创建：HEARTBEAT第12项执行中发现 chromadb/lancedb 在 venv 不在系统Python
