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 pm-done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。
正在显示 SKILL.md
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
| name | pm-done |
| description | Mark task done + schedule follow-up |
Mark a task as done and optionally create a follow-up task
| What | Path |
|---|---|
| Tasks | $PM_PATH/pm_tasks_master.csv |
| Projects | $PM_PATH/pm_projects_master.csv |
import pandas as pd
from datetime import date, timedelta
tasks = pd.read_csv('$PM_PATH/pm_tasks_master.csv', on_bad_lines='skip')
# Find by keyword, task_id, or project_id
# Example: find Client D invoice task
match = tasks[tasks['task_name'].str.contains('invoice', case=False, na=False) &
(tasks['status'] != 'done')]
today = str(date.today())
tasks.loc[tasks['task_id'] == 'task-XXX', 'status'] = 'done'
tasks.loc[tasks['task_id'] == 'task-XXX', 'last_updated'] = today
tasks.loc[tasks['task_id'] == 'task-XXX', 'notes'] = tasks.loc[
tasks['task_id'] == 'task-XXX', 'notes'
].fillna('').astype(str) + f' DONE {today}. [completion details]'
tasks.to_csv('$PM_PATH/pm_tasks_master.csv', index=False)
After marking done, ALWAYS ask user:
"Task marked done. Should I create a follow-up? For example: check payment in 5 days, verify delivery, etc."
If user says yes (or if the task type naturally needs a follow-up):
import uuid
followup_task = {
'task_id': f'task-{uuid.uuid4().hex[:4]}',
'project_id': original_task['project_id'],
'parent_task_id': '',
'task_name': 'Follow-up: [what to check]',
'description': '[Details of what to verify]',
'status': 'todo',
'priority': 'medium',
'priority_score': 0.5,
'assignee': 'Ivan',
'created_date': today,
'last_updated': today,
'deadline': str(date.today() + timedelta(days=N)), # typically 3-7 days
'estimated_hours': 0.25,
'actual_hours': '',
'actual_tokens': '',
'blocked_by': '',
'blocking': '',
'crm_activity_id': '',
'crm_person_linkedin_url': '',
'tags': original_task.get('tags', ''),
'notes': f'Follow-up from {original_task["task_id"]}',
'order_index': ''
}
tasks = pd.concat([tasks, pd.DataFrame([followup_task])], ignore_index=)
tasks.to_csv(, index=)
These task types ALWAYS need a follow-up (suggest automatically):
| Task type | Follow-up | Days |
|---|---|---|
| Invoice sent | Check if paid | 5-7 |
| Email sent (important) | Check for reply | 3-5 |
| Proposal/quote sent | Check for response | 3-5 |
| Deliverable submitted | Check for feedback | 3-5 |
| Meeting scheduled | Prepare for meeting | 1 day before |
| Escalation sent | Check for resolution | 2-3 |
IMPORTANT: When Claude helps execute a task during a conversation (sends email, creates invoice, writes a message, etc.), Claude MUST:
This prevents stale tasks from showing up in daily briefings.
pm_tasks_master.csv is for dev/project tasks only. Sales follow-ups (invoice checks, payment follow-ups, deal closing) are tracked in leads.csv via next_action and next_action_date. Do NOT create pm tasks for sales activities.
show-today -- daily task view (references this skill)daily-briefing -- morning summarycreate-project -- create project with taskslog-activity -- CRM activity loggingupdate-lead -- update sales follow-ups in leads.csv