用 Codex 或 Claude 帮你安装 复制这段 Prompt，粘贴到 Codex、Claude 或其他助手里，让它检查 Skill 页面并帮你完成安装。
直接命令不会经过审查 Prompt；运行前请先检查来源。
npx skills add https://github.com/aAAaqwq/AGI-Super-Team --skill channel-truth-run命令会保持在同一行。复制前请横向滚动并检查完整内容。
想先保存到本地？可下载 SkillsMP 当前能够提供的文件。
币安广场合约投机雷达 v5：以最近24小时专业交易帖为主要证据，回源核验帖子， 联合币安公共合约行情、4周期K线、布林带、ATR、量能和RR，生成可审计的本地影子报告。 触发词：币安广场、扫描币安、binance square、合约机会、交易信号雷达、4小时雷达
BTC 5分钟K线实时方向预测。v5.9对抗式审查重构: 13因子收敛到3个有证据信号(half_body延续+volume放量+meanrev回归, 11个47-49%硬币因子清零) + 三层独立信息过滤(多周期4h/1h/15m趋势 + 跨资产ETH/SOL广度 + 真订单流OFI) + 移除bull×0.92惩罚/Platt置信度门控。半K线策略第2分钟执行。黑天鹅防护: ATR spike+FNG<25。Binance端点双向故障切换。
BB 双向套利策略：加密合约 10x 杠杆布林带均值回归。布林带收窄=横盘→在下轨买、上轨卖；三重过滤器(1h趋势/RSI/BB甜区)确认碗平放，轨对轨止盈(RR 2:1~4:1)。含实时WebSocket模拟盘(paper)、历史回测(simulate/backtest_daily)、币安永续实盘CLI(trade_exec)。触发：'bb套利'、'布林带'、'bollinger'、'横盘策略'、'NEAR'、'回测'、'模拟盘'、'paper trading'。
基于 SOC 职业分类
正在显示 SKILL.md
| name | channel-truth-run |
| description | Automatic channel and contact synchronization |
Runs the Channel Truth Agent to synchronize last_contact timestamps
Channel Truth Agent:
activities.csvpeople.csv, leads.csv, clients.csv with current dates# Full run (default)
cd $AGENTS_PATH/channel-truth
python3 channel_truth_agent.py
# Dry-run (read only, no writes)
python3 channel_truth_agent.py --dry-run
# Single contact (for debugging)
python3 channel_truth_agent.py --person-id=p-acme-001
# Without channel scanning (activities.csv only)
python3 channel_truth_agent.py --no-channels
# Check if activities.csv exists
ls -lh $CRM_PATH/activities.csv
# Check people.csv
ls -lh $CRM_PATH/contacts/people.csv
# Dry-run test
python3 channel_truth_agent.py --dry-run
# View report
ls -lt $AGENTS_PATH/reports/channel_truth_*.md | head -1
# View latest report
cat $(ls -t $AGENTS_PATH/reports/channel_truth_*.md | head -1)
# View logs
tail -20 $AGENTS_PATH/logs/channel_truth.json
Updated CSV files:
people.csv - added last_contact and last_updatedleads.csv - added last_contact_via_primaryclients.csv - added last_contact_via_primaryReport:
$AGENTS_PATH/reports/channel_truth_YYYY-MM-DD.mdBackup files:
people.csv.bak.TIMESTAMPleads.csv.bak.TIMESTAMPclients.csv.bak.TIMESTAMPTelegram notification:
If something went wrong:
# Rollback via backup
cd $CRM_PATH/contacts/
ls -lt people.csv.bak.* | head -1
cp people.csv.bak.YYYYMMDD_HHMMSS people.csv
cd ../relationships/
cp leads.csv.bak.YYYYMMDD_HHMMSS leads.csv
cp clients.csv.bak.YYYYMMDD_HHMMSS clients.csv
# Or via git
cd $PROJECT_ROOT
git log --oneline sales/crm/
git revert <commit-hash>
Agent runs automatically via launchd:
~/Library/LaunchAgents/com.yourcompany.channel-truth-agent.plist# Check status
launchctl list | grep channel-truth
# Reload
launchctl unload ~/Library/LaunchAgents/com.yourcompany.channel-truth-agent.plist
launchctl load ~/Library/LaunchAgents/com.yourcompany.channel-truth-agent.plist
# Check path
ls -l $CRM_PATH/activities.csv
# If missing - create a basic one
echo "activity_id,person_id,company_id,product_id,type,channel,direction,subject,notes,date,created_by" > activities.csv
# Check Telegram credentials
ls -l $SALES_PATH/telegram/sessions/
# Test tg-tools
cd $TG_TOOLS_PATH
python3 -c "from tg_utils.auth import get_client; import asyncio; asyncio.run(get_client())"
# Skip channel scanning
python3 channel_truth_agent.py --no-channels
log-activity - logs activities in activities.csvgit-workflow - for committing changes (if needed)telegram-send - for manually sending messages# User: "run channel truth in dry-run"
cd $AGENTS_PATH/channel-truth
python3 channel_truth_agent.py --dry-run
# View what will be changed
cat /tmp/channel-truth-output.txt
# If everything is ok - run for real
python3 channel_truth_agent.py
# View report
cat $(ls -t $AGENTS_PATH/reports/channel_truth_*.md | head -1)