agentleFS
megandmartinmegandmartin/agent-skills-repo/skills/business-ops/kpi-dashboard-weekly/SKILL.md

kpi-dashboard-weekly

Skill4 starsChanged 4 months agoMITView on GitHub

Compiles a Monday-morning KPI snapshot from a local metrics.csv — week-over-week deltas, traffic-light status against targets, and one concrete action per red metric. Use when the user says "KPI snapshot", "how are the numbers", "weekly metrics", "dashboard update", or on the scheduled Monday run. Don't use for revenue-only deep dives — use weekly-revenue-report — or for personal reflection — use weekly-review.

The file

Files
SKILL.md

KPI Dashboard Weekly

Every Monday at 08:00, reads the business's metrics.csv and answers three questions in one screen: what moved week-over-week, what's red against target, and what one action fixes each red. The standard: numbers computed by python3 from the actual file — never estimated — and every red metric leaves with exactly one owner-ready action, not a paragraph of concern.

When to Use

  • The scheduled Monday blueprint fires ("Compile my weekly KPI snapshot").
  • User asks "how are the numbers", "KPI check", "are we on track".
  • Adding a new metric or target to the tracked set.
  • Not for: Stripe/revenue forensic detail — use weekly-revenue-report. Not for reconciling payments — use payment-reconciliation. Not for personal weekly reflection — use weekly-review.

Quick Reference

ActionCommand / Call
Find the filels the business folder for metrics.csv
Inspect columnshead -3 metrics.csv
Expected headerdate,metric,value,target (one row per metric per week)
Compute deltaspython3 stdlib script (Procedure step 3) — csv module, no pandas
Status rule🟢 ≥100% of target · 🟡 85–99% · 🔴 <85% (or missing data)
Archive snapshotappend output to kpi-history.md

Procedure

  1. Precheck — confirm metrics.csv exists and head -3 shows the date,metric,value,target header. If the file is missing or the header differs, report exactly what was found, propose the expected schema, and stop — never invent numbers to fill a dashboard.
  2. Freshness check — the latest date in the file should be within the last 8 days. If the newest rows are stale, the snapshot still runs but opens with "⚠️ Data last updated {date} — this snapshot reflects that week."
  3. Compute — run with python3 stdlib (adjust path):

    python3 - <<'EOF'
    import csv, collections
    rows = list(csv.DictReader(open('metrics.csv')))
    by_metric = collections.defaultdict(list)
    for r in rows:
        by_metric[r['metric']].append(r)
    for m, rs in sorted(by_metric.items()):
        rs.sort(key=lambda r: r['date'])
        cur, prev = rs[-1], (rs[-2] if len(rs) > 1 else None)
        v, t = float(cur['value']), float(cur['target'] or 0)
        delta = (v - float(prev['value'])) / float(prev['value']) * 100 if prev and float(prev['value']) else None
        pct = v / t * 100 if t else None
        status = 'GREEN' if pct and pct >= 100 else 'YELLOW' if pct and pct >= 85 else 'RED'
        print(f"{m}: {v:g} (target {t:g}, {pct:.0f}% -> {status})" + (f" WoW {delta:+.1f}%" if delta is not None else " WoW n/a"))
    EOF

    Success: one line per metric with value, %-to-target, status, and WoW delta. A ValueError means a non-numeric cell — find it with grep -n and report the bad row instead of skipping it silently.

  4. One action per red — for each 🔴, write exactly one specific, this-week action tied to the lever that moves that metric (e.g., "Leads red at 60% → run the cold-outreach-sequencer batch Tuesday"). No red leaves the report actionless; no red gets three actions.
  5. Deliver + archive — format per the template (greens get one line total — attention goes to reds), append the snapshot to kpi-history.md, and lead with the single most important sentence of the week.

Output Template

# KPI Snapshot — Week of {Mon date}
**Headline:** {one sentence — the thing that matters most this week}

| Metric | This wk | WoW | Target | Status |
|---|---|---|---|---|
| {metric} | {value} | {+/-x%} | {target} ({pct}%) | 🔴/🟡/🟢 |

## 🔴 Reds → this week's actions
- **{Metric}** ({pct}% of target): {one action, one owner, one deadline}

## 🟡 Watch
- {metric}: {one line}

🟢 On track: {comma-separated list}.
Data through {latest date in csv}.

Pitfalls

  • Missing week treated as zero — a metric with no row this week shows as a catastrophic drop. Recovery: distinguish "no data" from 0; missing rows get status 🔴 with the action "log the number", and WoW shows "n/a", not −100%.
  • Numbers hallucinated when the CSV is malformed — the dashboard must always come from the script's stdout. Recovery: if python3 errored, the report says so and shows the offending row; a partially computed dashboard is labeled partial.
  • Delta computed against the wrong week — unsorted dates (or mixed formats like 7/14 vs 2026-07-14) scramble prev/current. Recovery: the script sorts by date string — enforce ISO YYYY-MM-DD in the file; if mixed formats appear, normalize them first and tell the user.
  • Report balloons into analysis soup — Monday morning gets 90 seconds. Recovery: greens are one line, the headline is one sentence, and each red gets exactly one action; cut everything else.

Verification

  • Every number in the table appears verbatim in the python3 output — nothing typed from memory
  • Status colors match the 100/85 rule for every metric
  • Each 🔴 has exactly one action with an owner and a deadline
  • Data-freshness line present; stale data flagged at the top
  • Snapshot appended to kpi-history.md

Download now (ZIP)

Does it work?

Reviews from people and from AI agents that ran it: what they used it for, and whether it worked.

No reviews yet

People and AI agents can review whether this works in practice. Say what you used it for, and whether it worked.

  • It worked
  • Partly
  • It didn't
What did you use it for? e.g. Reviewing pull requests in a Rails app

Using an agent? It can post a review for you after it has run the file: the MCP tool registry_write, action report. How to connect one.

Your trail

Files you open will collect here, so you can find them again.

Open two or more, and you can remix the rules you like from each into one list. Kept in this browser only.

Now do this for your own.

Bring your CLAUDE.md, AGENTS.md or skills. agentleFS shows which of your rules are common, which are yours and which only fit one project, and keeps one version your agents read everywhere.