daily-pulse
GitHub为小企业主生成每日业务快照，涵盖销售、人力成本、COGS及P&L快报。通过对比目标与历史数据，自动标记异常指标，帮助经营者在问题恶化前掌握经营状况，实现24小时内的快速决策。
Trigger Scenarios
Install
npx skills add cosmicstack-labs/mercury-agent-skills --skill daily-pulse -g -y
SKILL.md
Frontmatter
{
"name": "daily-pulse",
"metadata": {
"tags": [
"daily-reporting",
"p-and-l",
"sales-tracking",
"kpi-dashboard",
"financials",
"restaurant-metrics",
"business-intelligence"
],
"author": "cosmicstack-labs",
"version": "1.0.0",
"category": "shop-restaurant"
},
"description": "Generate a daily business snapshot — sales, labor cost, COGS, covers, and P&L flash report — so the owner knows exactly where the business stands every morning. Flags anomalies before they become problems."
}
Daily Pulse
Core Principles
The difference between a thriving shop and a failing one is often simply knowing the numbers daily. Most small business owners fly blind until the monthly P&L arrives 3 weeks after month-end — too late to fix anything. The Daily Pulse skill compresses the P&L cycle to 24 hours, giving the owner a crystal-clear morning snapshot of yesterday's performance.
The Five Daily Numbers That Matter
- Total Sales — Yesterday's revenue vs. target vs. same day last year
- Covers / Transactions — How many customers (for restaurants) or transactions (for retail)
- Average Check / Ticket — Revenue ÷ covers. Trending up or down?
- Labor Cost % — Total labor ÷ sales. Warning if > budget.
- COGS % — Cost of goods sold ÷ sales (restaurants) or COGS as % of sales (retail)
Skill Workflow
Step 1 — Connect Data Sources
Typical data sources (ask user which they use):
- POS System: Square, Toast, Clover, Lightspeed, Shopify
- Payroll System: Gusto, ADP, or manual hours
- Inventory System: Current stock counts, received orders
- Bank Account: For actual COGS (invoiced amounts)
If no system integration is available:
"No problem. I can build your daily report from a simple template. Each evening, answer 5 quick questions and I'll generate your pulse report. Let me set that up."
Step 2 — Build the Daily Flash Report
Generate a structured daily summary:
📊 DAILY PULSE — Tuesday, March 11, 2025
━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━
💰 SALES
Yesterday: $3,840
Target: $4,000 (96% of target)
Same Day Last Week: $3,620 (+6.1% vs. last week)
Same Day Last Year: $3,150 (+21.9% vs. last year)
👥 CUSTOMERS
Covers/Transactions: 172
Avg Check: $22.33 (vs. $21.05 last week — +6.1%)
Peak Hour: 7:00-8:00pm (38 covers)
👷 LABOR
Hours Scheduled: 84 hrs
Hours Worked: 82 hrs
Labor Cost: $1,476
Labor %: 38.4% (Budget: 35%) ⚠️ ABOVE TARGET
Overtime Used: 2 hrs ($54)
🥩 COGS (estimates)
Today's COGS: $1,152 (30% of sales)
MTD COGS %: 30.5% (Target: 28-32%) ✅ ON TRACK
📈 MTD PERFORMANCE
Month-to-Date Sales: $42,560
MTD Target: $44,000 (96.7% of target)
MTD Labor %: 36.2% (Target: 35%) ⚠️ Slightly over
MTD COGS %: 30.5% ✅ On target
Projected Monthly: $127,680 (if current pace holds)
Step 3 — Flag Anomalies & Alerts
Compare each metric against thresholds and flag variances:
| Alert Level | Condition | Example |
|---|---|---|
| 🔴 Critical | Sales < 80% target OR Labor % > 45% | "Sales dropped 30% vs last Tuesday — check for nearby event or competitor opening" |
| 🟠 Warning | Sales 80-90% target OR Labor % 40-45% OR COGS > 35% | "Labor % at 38.4%. You're $94 over budget for the day" |
| 🟢 Good | All metrics within target range | "Solid day. Avg check ticked up — menu engineering is working" |
| ⭐ Notable | Any metric significantly positive | "Saturday smashed target by 24% — best day this quarter. Well done!" |
Generate specific, actionable alerts:
⚠️ Labor Alert: Labor % at 38.4% vs. 35% target. You scheduled 84 hours but used 2 hours overtime. Consider splitting the close shift earlier to reduce OT.
📉 Sales Alert: Tuesday sales were 96% of target — close but missing pattern. Wednesdays have averaged 92% of target for 3 weeks. Consider a Wednesday special to lift midweek traffic.
Step 4 — Rolling Week & Month Trends
Provide trend context:
📈 7-DAY TREND
Sales: ↑↑ (up 8% vs prior 7 days)
Covers: ↑ (up 3%)
Avg Check: ↑↑ (up 5% — menu price increase flowing through)
Labor %: → (flat at 36.5%)
COGS %: ↓ (down 1.5% — portion control improving)
📊 MONTH-TO-DATE vs. BUDGET
Total Sales: $42,560 / $44,000 (97%)
Total Labor: $15,407 / $15,400 (~100%)
Total COGS: $12,981 / $13,200 (98%)
Net Margin: 21.3% (estimated) vs. 22% target
Step 5 — Action Recommendations
Close the report with 1-3 action items for the day:
Today's action items:
- 🔴 Review Friday's overtime. Two employees exceeded 40 hrs — consider midweek day off for them.
- 🟠 Wednesday lunch covers have dropped for 3 weeks. Test a $9.99 lunch special.
- 🟢 Keep doing what you're doing with the new menu pricing — avg check is up 5%.
Step 6 — Schedule Daily Delivery
"I can deliver this report to you every morning by 8am. Would you like it as a summary here, or shall I send it to your email/phone? Also, I can include a simple 5-question evening prompt so you can log the data if your systems don't auto-connect."
Trigger Phrases
| Phrase | What Happens |
|---|---|
| "How did we do yesterday?" | Generates the daily pulse report |
| "Give me today's numbers" | Same — daily summary snapshot |
| "Run the daily report" | Full flash report with all KPIs |
| "Any red flags?" | Shows only alerts and anomalies |
| "How's my month looking?" | MTD performance vs. budget projection |
| "Compare this week to last week" | 7-day trend analysis |
| "What should I focus on today?" | Action item recommendations based on data |
Key Instructions Summary
- Collect data: Sales, covers, labor hours, inventory usage from POS/payroll/user input
- Build report: Structured flash report with all key metrics, comparisons, and percentages
- Flag anomalies: Red/orange/green alerts with specific commentary
- Show trends: 7-day rolling and MTD context
- Recommend actions: 1-3 concrete action items for the day ahead
- Schedule: Offer daily automated delivery at a set time
Common Mistakes
- Looking at sales in isolation. Revenue of $5k sounds great — but if labor was $2k and COGS was $2k, you made nothing. Always look at the margin stack.
- Comparing to arbitrary targets. "We did $4k yesterday — is that good?" Compare to budget, same day last week, and same day last year.
- Ignoring trends for single-day spikes. One bad Monday doesn't mean doom. One bad month does. Watch rolling averages.
- Not acting on the data. A daily report is useless if it doesn't change behavior. Every alert should have a recommended action.
- Overcomplicating it. 5 key metrics, reviewed daily, beats 50 metrics reviewed monthly. Start simple.
"What gets measured gets managed. What gets measured daily gets improved daily."
Version History
- 38e2523 Current 2026-07-05 19:41
