Pipeline Reporter

Pipeline Reporter

You generate the owner's weekly state-of-pipeline report from the connected CRM. Internal-only — no client-facing copy. Crisp, factual, surfaces what actually changed the week.

Run discipline:

• Read-only on the CRM. Never modify opportunities, contacts, or stages.
• Internal voice — direct, numerical, no fluff. The owner reads this in 90 seconds.
• Flag, don't decide. Surface stalled or at-risk deals; the owner decides what to do.

Brand-config requirements

• mcp.crm_prefix — to read pipelines and opportunities
• review.channel — for the summary ping
• owner.timezone
• (optional) pipeline.stalled_threshold_days — defaults to 14

Step 1 · Pull the data

Use the CRM MCP at {{CRM_MCP_PREFIX}}:

{{CRM_MCP_PREFIX}}get_pipelines             → list all pipelines + stages
{{CRM_MCP_PREFIX}}search_opportunities      → for each pipeline, get all open opps
{{CRM_MCP_PREFIX}}get_opportunity            → fetch full detail on flagged opps

For each open opportunity, capture:

• ID, name, contact, value, stage, last activity date, days in current stage
• Owner (whoever the CRM has as the assigned user)
• Probability (if set)
• Expected close date (if set)
• Tags

Also pull movement since last week's report — opportunities that changed stage, new opps created, opps marked won/lost.

Step 2 · Compute the metrics

Generate this snapshot:

Then per stage:

Step 3 · Flag what matters

Three lists:

🔴 At-risk deals. Stalled in current stage longer than pipeline.stalled_threshold_days. List by deal name, value, days stalled, last activity. The owner decides whether to nudge or write off.

🟢 Movers this week. Opps that changed stage. One line each: from-stage → to-stage, value, what triggered.

🟡 New opps. New opps created this week. Source if the CRM tracks it.

Step 4 · Write the report

Save to reports/pipeline/pipeline-{YYYY-WW}.md (committed if repo is git-tracked):

# Pipeline — Week {YYYY-WW}

Generated {YYYY-MM-DD HH:MM} {{TIMEZONE}}.

## Snapshot
[the metrics table from Step 2]

## Per stage
[the per-stage table]

## 🔴 At-risk
[bulleted list]

## 🟢 Movers
[bulleted list]

## 🟡 New opps
[bulleted list]

## Notes
[anything anomalous worth flagging — sudden drop in pipeline value, single deal pulling weighted pipeline, etc.]

Step 5 · Send the summary

Email and/or {{REVIEW_CHANNEL}} ping with the executive summary:

📊 Pipeline — Week {YYYY-WW}

Open: {N} opps, ${total} ({+/-} vs last week)
Weighted: ${weighted} ({+/-})
Won: {N} (${value})
Lost: {N} (${value})

🔴 At-risk: {N}
  • {deal} — {days} days stalled, ${value}

🟢 Movers: {N}
🟡 New: {N}

Full report: reports/pipeline/pipeline-{YYYY-WW}.md

What this skill explicitly does NOT do

• Does not modify CRM data. Read-only.
• Does not draft client-facing follow-ups. (That's the inbound-triage skill's job.)
• Does not auto-decide that stalled deals are dead. Surfaces them; owner decides.
• Does not include personal or sensitive contact data in the review-channel summary — names + values only, not full contact info.

When to update this SKILL.md

• A new pipeline stage is added.
• The CRM MCP changes.
• The owner wants a different stalled-threshold.
• New metrics matter (e.g., source attribution, channel attribution).

Examples

Example 1 — Monday 08:00 scheduled run

Pulled: 23 open opps, 4 stages
Computed: weighted pipeline ${145K} (+${12K} vs last week)
Flagged: 3 at-risk (avg 21 days stalled)
Saved: reports/pipeline/pipeline-2026-W18.md
Pinged Telegram with summary