Customer Service Operations Skill

AI-powered customer service — KB-first resolution, churn prediction, intelligent routing, and satisfaction tracking. Triage → Understand → Resolve → Respond → Close.

Skill Standard MCP Server ADK-Rust Enterprise License

What This Skill Does

This skill turns the 20-tool mcp-customer-service server into a context-aware support operator that checks customer health before responding, resolves from KB first, and proactively flags churn risks.

Workflow Tool Calls What It Achieves
Handle Conversation 6-7 Full triage → understand → resolve → respond loop
Churn Prevention 3 Risk assessment with actionable retention signals
Queue Management 2-3 Queue health with agent capacity and SLA risks
Service Metrics 2 CSAT, NPS, FCR, response time dashboard
Routing & Escalation 2-3 Intelligent assignment with context handoff

The Core Loop

Triage (what's urgent?) → Understand (who is this customer?)
    → Resolve (KB first!) → Respond (personalized) → Close (CSAT triggered)

Key Differentiators:

  • Customer health drives tone — declining health = extra empathy
  • KB-first resolution — 30%+ issues resolved without custom responses
  • Churn-aware — every interaction checks risk signals
  • VIP treatment — Enterprise customers auto-routed to senior agents

Installation

# Claude Code
git clone https://github.com/zavora-ai/skill-customer-service-operations.git \
  ~/.skills/skills/customer-service-operations

# ADK-Rust
cp -r customer-service-operations /path/to/project/.skills/skills/

Requirements

Required:

  • mcp-customer-service server connected (v2.1.0+)

Optional:

  • mcp-crm — enrich with deal context and CRM activities
  • mcp-slack — escalation alerts and team notifications
  • mcp-email — proactive outreach for churn prevention
  • mcp-notifications — SLA breach alerts to agents

Folder Structure

customer-service-operations/
├── SKILL.md                        # Main skill (loaded on trigger)
├── scripts/
│   └── churn_score.py              # Composite churn risk calculator
├── assets/
│   ├── queue-status-report.md      # Queue health template
│   └── service-metrics-report.md   # CSAT/NPS/FCR dashboard template
├── references/
│   ├── tool-sequences.md           # 20 tools + priority matrix
│   ├── cross-mcp-workflows.md      # CS + CRM + Slack + Email
│   └── examples.md                 # 4 real-world scenarios
├── README.md
└── LICENSE

Example

User: "Handle the urgent billing complaint"

Agent behavior:

  1. Lists urgent open conversations
  2. Reads full thread + customer profile + health score
  3. Assesses churn risk (high: 82/100)
  4. Searches KB for "duplicate charge" resolution
  5. Generates personalized response with empathy (health declining)
  6. Replies + adds internal note flagging churn risk

Result:

✅ Handled conv-3: Billing duplicate charge
Customer: Tom Wilson | Health: 30 | Churn Risk: 🚨 HIGH
Resolution: Refund initiated per KB article #12
⚠️ Recommendation: Proactive retention outreach needed

Success Criteria

Metric Target
KB resolution rate 30%+ resolved from knowledge base
Response quality Context-aware (checks health before responding)
Churn detection Flags high-risk customers proactively
Trigger rate 90% on support-related queries

MCP Server Compatibility

Designed for mcp-customer-service v2.1.0+:

Capability Tools
Conversations list, get, start, reply, note (5)
Customer Intelligence profile, health, history, churn (4)
Resolution KB search, suggest, canned, resolve (4)
Routing assign, escalate, queue status (3)
Metrics CSAT/NPS, service metrics, agents, merge (4)

Contributors

James Karanja Maina

License

Apache-2.0

Part of the ADK-Rust Enterprise skills ecosystem.

Built with ❤️ by Zavora AI