Overview

MCP prompts are pre-built conversation templates that fetch live data from ProBeya (via tRPC), assemble it into a well-structured context block, and return it as a message the AI can analyze. They enforce consistent analysis structure for recurring pharma operational workflows.

All prompts accept arguments via their argsSchema (Zod-validated) and return a single user role message with formatted text. The AI then analyzes the data and provides structured recommendations.

When to use prompts vs tools:

  • Tools retrieve or modify a single piece of data (list KPIs, create an action)
  • Prompts orchestrate multi-step data retrieval and request structured analysis (SQCDP review, CAPA documentation)

daily_standup

Title: Daily Standup Summary

Generate a daily standup briefing for a board, highlighting blockers, KPI alerts, overdue actions, and priorities for the day. Designed for 5-minute AIC/TIER 1 morning meetings.

PropertyValue
ArgumentboardId (string, required)

Data fetched automatically:

  • Board data (item count, group distribution) via boards.getByProjectId
  • KPI alerts (red/amber breaches) via kpis.getAlerts
  • Overdue actions via actions.list with overdue: true

Invoking the Prompt

In a conversation, ask:

“Run the daily standup prompt for the Brussels filling line board”

Or use the prompt selector (/) and choose daily_standup.

Example Output

The AI receives the following context and produces a structured briefing:

DAILY STANDUP BRIEFING — Brussels Filling Line
═══════════════════════════════════════════════

KPI ALERTS: 2 active
━━━━━━━━━━━━━━━━━━━
1. OEE — 82.5% (AMBER) | Target: 90% | Green: ≥85%
   Root cause: Line 2 seal failure, 45min unplanned downtime
   → Maintenance scheduled for tonight's shift

2. Right First Time — 96.2% (AMBER) | Target: 99% | Green: ≥98%
   Root cause: 3 labeling rejects on secondary packaging
   → Not a product quality issue; label printer alignment checked

OVERDUE ACTIONS: 3
━━━━━━━━━━━━━━━━━
1. [CRITICAL] CAPA-2026-0187: Replace conveyor belt section 3
   Owner: Dr. Chen | Due: Apr 5 (8 days overdue) | Tier 1
   → Parts arrived yesterday. Scheduled for installation tonight.

2. [HIGH] Review SOP-MFG-042 temperature monitoring procedure
   Owner: J. Dubois | Due: Apr 10 (3 days overdue) | Tier 1
   → Draft complete, pending QA review.

3. [MEDIUM] Update training records for new filling operators
   Owner: S. Van Damme | Due: Apr 11 (2 days overdue) | Tier 1
   → 3 of 5 operators completed. Remaining 2 scheduled this week.

TODAY'S PRIORITIES:
1. Complete CAPA-0187 conveyor belt installation (safety critical)
2. Get QA sign-off on SOP-MFG-042 revision
3. Monitor OEE recovery after seal replacement
4. Close labeling issue on secondary packaging line

board_analysis

Title: Board Health Analysis

Analyze a board’s item distribution, identify bottlenecks, assess workload balance, and recommend flow improvements.

PropertyValue
ArgumentprojectId (string, required)

Data fetched automatically:

  • Full board data (groups, columns, items, values) via boards.getByProjectId
  • Group-level item counts computed from board data

Analysis the AI performs:

  1. Bottleneck identification — which groups have disproportionate item counts?
  2. Workload balance — are items evenly distributed across assignees?
  3. Column usage patterns — are all custom fields being populated?
  4. Flow improvement recommendations — WIP limits, swimlane restructuring

Example Output

BOARD HEALTH ANALYSIS — CAPA Tracking
══════════════════════════════════════

BOTTLENECK: "In Review" group has 14 items (47% of total)
  Average time in "In Review": 8.3 days
  → Recommendation: Add a WIP limit of 5. Assign a dedicated QA reviewer
    for weekly CAPA review sessions.

WORKLOAD IMBALANCE:
  Dr. Chen: 9 items (30%)  ← Overloaded
  J. Dubois: 6 items (20%)
  S. Van Damme: 3 items (10%)
  Unassigned: 12 items (40%) ← High unassigned rate
  → Recommendation: Redistribute 3 items from Dr. Chen. Assign the
    12 unassigned items during the next tier meeting.

COLUMN USAGE:
  "Due Date" populated: 73% (8 items missing due dates)
  "Priority" populated: 100%
  "Root Cause Category" populated: 45% ← Low
  → Recommendation: Make "Root Cause Category" required. Missing
    categorization prevents trend analysis.

escalation_report

Title: Tier Escalation Report

Summarize all actions across tiers for management review. Designed for bi-weekly site director meetings and monthly executive reviews.

PropertyValue
ArgumentboardId (string, required)

Data fetched automatically:

  • Tier 1 (Team) actions via actions.list with tierLevel: 1
  • Tier 2 (Department) actions via actions.list with tierLevel: 2
  • Tier 3 (Site) actions via actions.list with tierLevel: 3

Analysis the AI performs:

  1. Critical escalations needing immediate attention
  2. Patterns across tiers (recurring categories, common root causes)
  3. Recommendations for de-escalation
  4. Root cause themes
  5. Executive summary for the site director

Example Output

TIER ESCALATION REPORT — Brussels Manufacturing Site
═════════════════════════════════════════════════════

EXECUTIVE SUMMARY:
23 total actions across 3 tiers. 2 critical escalations require
immediate site director attention.

TIER 3 — SITE (2 actions)
━━━━━━━━━━━━━━━━━━━━━━━━
[CRITICAL] Capital request: Replace aging filling line isolator
  Escalated from T2 on Apr 8. Estimated cost: EUR 450K.
  Impact: Current isolator has 3 integrity failures in 6 months.
  Recommendation: Fast-track capital approval. Patient safety risk.

[HIGH] Cross-department API supply chain contingency
  Escalated from T2 on Apr 5. Single-source API from India.
  Recommendation: Qualify backup supplier. 9-month lead time.

TIER 2 — DEPARTMENT (7 actions)
━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━
  Quality: 4 actions (3 overdue)
  Delivery: 2 actions
  Safety: 1 action
  → PATTERN: Quality actions dominate T2. Root cause: insufficient
    QA reviewer capacity. Recommend hiring additional QC analyst.

TIER 1 — TEAM (14 actions)
━━━━━━━━━━━━━━━━━━━━━━━━━━
  Open: 3 | Assigned: 5 | In Progress: 4 | Blocked: 2
  → 2 blocked actions need T2 escalation (waiting on parts procurement)

kpi_review

Title: Weekly KPI Review (SQCDP)

Structured SQCDP category review with trend analysis and recommended corrective actions. Designed for weekly management review meetings.

PropertyValue
ArgumentboardId (string, required)

Data fetched automatically:

  • All KPI definitions with latest values via kpis.list
  • KPI alerts (red/amber breaches) via kpis.getAlerts

Analysis per SQCDP category:

  1. Current status (green/amber/red)
  2. Trend direction (improving/stable/degrading)
  3. Root cause for any deviations
  4. Recommended corrective actions
  5. Impact on pharma operations (batch release, compliance, patient safety)

Example Output

WEEKLY SQCDP KPI REVIEW — Brussels Filling Line
════════════════════════════════════════════════
Week of April 7-13, 2026

SAFETY [GREEN] ✓
  LTI Rate: 0 (target: 0) — 847 days without LTI
  Near-misses: 2 reported (target: >5 reports/week)
  → Trend: Stable. Near-miss reporting below target — may indicate
    under-reporting rather than good performance. Recommend safety
    observation blitz this week.

QUALITY [AMBER] ⚠
  Right First Time: 96.2% (target: 99%, threshold: 98%)
  Batch Rejection Rate: 0.8% (target: <0.5%)
  → Trend: Degrading. 3rd consecutive week below target.
  → Root cause: Labeling accuracy on secondary packaging.
  → Corrective action: Replace label printer on Line 2 (ordered).
  → GxP impact: No product quality impact. Cosmetic defect only.

COST [GREEN] ✓
  Cost per unit: EUR 2.34 (target: EUR 2.50)
  Scrap rate: 0.3% (target: <1%)

DELIVERY [AMBER] ⚠
  OEE: 82.5% (target: 90%, threshold: 85%)
  OTIF: 94% (target: 98%)
  → Trend: Improving (was 79.1% last week).
  → Root cause: Unplanned downtime on Line 2 (seal failure).
  → Corrective action: CAPA-0187 in progress. Seal replaced.
  → GxP impact: 2 batch deliveries delayed by 24 hours.

PEOPLE [GREEN] ✓
  Absenteeism: 3.2% (target: <5%)
  Training completion: 87% (target: 95%)
  → Training gap: 5 operators pending GMP refresher course.

OVERALL HEALTH SCORE: 7/10
TOP 3 PRIORITIES:
1. Close quality gap — label printer replacement (due Apr 15)
2. Sustain OEE recovery — monitor Line 2 after seal replacement
3. Complete operator GMP training (5 remaining)

capa_analysis

Title: CAPA Analysis (Corrective and Preventive Action)

Analyze overdue and blocked actions for CAPA documentation suitable for quality management system (QMS) records and regulatory compliance (GxP).

PropertyValue
ArgumentboardId (string, required)

Data fetched automatically:

  • Overdue actions via actions.list with overdue: true
  • Blocked actions via actions.list with status: "blocked"

Analysis per issue — full CAPA documentation:

  1. Issue Description — what happened, when, severity
  2. Root Cause Analysis — 5 Whys or Ishikawa categories
  3. Corrective Action — immediate fix with responsible person and deadline
  4. Preventive Action — systemic change to prevent recurrence
  5. Effectiveness Check — how to verify the CAPA worked
  6. GxP Impact Assessment — impact on product quality, patient safety, data integrity

Example Output

CAPA ANALYSIS REPORT — Brussels Filling Line
═════════════════════════════════════════════
Generated: 2026-04-13 | For QMS record insertion

CAPA-2026-0194: Batch BX-4471 OOS — pH Excursion
━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━

ISSUE DESCRIPTION:
  What: pH reading 7.8 on Batch BX-4471 (spec: 7.0-7.4)
  When: 2026-04-06 during in-process testing
  Severity: Major — batch held pending investigation
  Impact: 1 batch affected, no patient exposure (batch not released)

ROOT CAUSE ANALYSIS (5 Whys):
  Why 1: Buffer pH was out of range
  Why 2: NaOH concentration in buffer preparation was incorrect
  Why 3: Operator used reagent-grade NaOH instead of USP-grade
  Why 4: Both grades stored in same cabinet with similar labels
  Why 5: SOP-WH-012 does not address reagent grade segregation
  ROOT CAUSE: Inadequate warehouse SOP for reagent grade segregation

CORRECTIVE ACTION:
  Action: Segregate NaOH grades into separate, labeled cabinets
  Responsible: J. Dubois (Warehouse Manager)
  Deadline: 2026-04-15
  Status: In progress — cabinets ordered, arriving Apr 14

PREVENTIVE ACTION:
  Action: Update SOP-WH-012 to require grade-specific storage locations
         with color-coded labels and barcode verification at dispensing
  Responsible: Dr. Chen (QA Manager)
  Deadline: 2026-04-25
  Training: All warehouse staff to be retrained on updated SOP

EFFECTIVENESS CHECK:
  Method: 30-day monitoring of buffer preparation records
  Metric: Zero reagent grade errors over 30 consecutive days
  Review date: 2026-05-25
  Reviewer: QA Director

GxP IMPACT ASSESSMENT:
  Product quality: No released product affected (batch held)
  Patient safety: No risk (batch not distributed)
  Data integrity: All records maintained; no ALCOA concern
  Regulatory: Potential FDA 483 observation if systemic issue identified
              Recommend voluntary disclosure in next annual report

Invoking the Prompt

“Run CAPA analysis on the Brussels filling line board”

Claude will invoke the capa_analysis prompt, fetch overdue and blocked actions, and produce a QMS-ready CAPA report for each issue.


Prompt Discovery

MCP clients can list all available prompts via the prompts/list endpoint:

{
  "method": "prompts/list",
  "params": {}
}

// Response
{
  "prompts": [
    { "name": "daily_standup", "description": "Daily Standup Summary" },
    { "name": "board_analysis", "description": "Board Health Analysis" },
    { "name": "escalation_report", "description": "Tier Escalation Report" },
    { "name": "kpi_review", "description": "Weekly KPI Review (SQCDP)" },
    { "name": "capa_analysis", "description": "CAPA Analysis (Corrective & Preventive Action)" }
  ]
}

Building Custom Prompts

The 5 built-in prompts cover the most common pharma workflows. For custom analysis patterns, combine tools in sequence:

User: Analyze our site's readiness for the upcoming FDA audit.

Claude: I'll gather the data from multiple sources.
  [list_compliance_requirements] → 47 requirements, 3 with gaps
  [list_inspection_checklists] → 5 checklists, 1 overdue session
  [list_actions with category=quality and overdue=true] → 6 overdue CAPAs
  [get_kpi_alerts] → 2 KPIs in amber state

Based on the data, here is your FDA audit readiness assessment:
  ...

This approach gives you the flexibility of custom analysis while leveraging ProBeya’s structured data through MCP tools.