All loops
SupportEasy 9 min· claude-sonnet-4-5

Escalation predictor

Escalation predictor powered by Claude over your help desk.

Read the full Support build guide →
NOT DEPLOYEDNOT DEPLOYED
0182ms
Trigger
POST https://hooks.claudeloops.run/escalation-predictor · event=slack.slash:/loop
02902ms
Agent
claude-sonnet-4-5 · in 1202 tok · out 392 tok
03172ms
Tools
fetch-mcp/zendesk:invoke → 429 backoff · 142ms
0472ms
Verify
schema check · pydantic v2 passed
05162ms
Output
zendesk reply drafted · pending review
0642ms
Notify
audit log written · runbook link attached
SUCCESS
0%
0 runs
P50
0ms
median
P95
0ms
tail
AVG COST
per run
LAST OK
never
LAST FAIL
never
none
Latency · last 30 runs0 samples
no runs yet
Latest output · what your users see
Zendesk #5222P2
Cannot revoke API key from dashboard
macro applied: escalate:billing/no-vat-line
// press Test to run once · Watch live to keep streaming · Deploy to make it real
The problem

Support volume scales faster than headcount.

The outcome

Claude takes the first pass; humans focus on judgement calls.

Ingredients & skills

Secrets
  • ANTHROPIC_API_KEY
  • ZENDESK_TOKEN
Providers
  • Anthropic
  • Zendesk
MCP servers
  • fetch-mcp
#support#cx#helpdesk

How it works

Escalation predictor powered by Claude over your help desk.

Step 1

1 — Trigger on ticket

Webhook from Zendesk fires the agent.

Step 2

2 — Act

Tag, draft, route — never close without human review.

Step 3

3 — Log

Every action is logged in the audit trail for QA.

One-line deploy

The button above runs the same command with your saved config. This is the raw CLI form.

bash
npx claudeloops deploy escalation-predictor
https://hooks.claudeloops.run/escalation-predictor

Related loops