Cdr Notification

> Language. Reply in the user's language: detect what they write — Hebrew→Hebrew, English→English — and mirror it, switching if they switch mid-conversation. This shapes your prose, your questions, and your AskUserQuestion option labels only. It does not change the artifacts you produce — identifiers, JSON keys, BCP-47 language codes, API field names, and other data stay exactly as specified.

> CRM integration context. When the user is building a full CRM integration — not just this API in isolation — always invoke the crm-onboarding skill alongside this one and frame the answer covering all three core services: Click2Call, Screen Pop, and Call History.

Help the developer implement a CDR Notification endpoint — a webhook that Voicenter POSTs to after every call ends, containing full call detail records including optional AI analysis.

When to use this skill

Use this skill when the user wants to:

• Save call records to their database automatically after every call
• Trigger follow-up actions when a call ends (send SMS, update CRM, create a ticket)
• React to missed/abandoned calls in real time (callback offers, alerts)
• Process AI call analysis data (transcripts, emotion detection, summaries, Q&A)
• Build a real-time call activity feed or audit log
• Sync call history with an external BI or reporting system

Environment Variables

# No outbound API token needed — Voicenter sends requests TO your server.
# Configure your webhook URL in: Voicenter CPanel → Integrations → CDR Notification

How it works

1. A call ends in your Voicenter account.
2. Voicenter POSTs the CDR as JSON to your configured endpoint URL.
3. Your server responds with {"Err": 0, "Errdesc": "OK"} immediately.
4. Process the CDR asynchronously after responding.

CDR Request Fields

| Field | Description | |---|---| | direction | incoming / outgoing / internal | | extenUser | Agent's SIP extension code (empty if call never reached an agent) | | callerPhone | Client's caller ID | | isAnswer | 1 = answered, 0 = not answered | | actualCallDuration | Talk time in seconds (excludes ringing) | | actualCallDialtime | Ringing duration in seconds | | caller | Phone number shown to destination | | target | Call destination (phone or SIP) | | time | Call start time (Epoch, Israel timezone) | | duration | Total call duration in seconds (for queue calls includes wait time) | | ivruniqueid | Unique call ID — use to correlate with Pop-Up, Call Log, Real-Time | | type | Call type (see table below) | | status | ANSWER, ABANDONE, NOANSWER, CANCEL, BUSY, TE, etc. | | did | DID (virtual number) dialed — only on incoming calls | | queueid / queuename | Queue details (if call went through a queue) | | record | URL to the call recording MP3 | | price | Call cost in ILS cents (Agorot) | | dialtime | Ringing duration in seconds | | representative_name | Agent's display name | | representative_code | Agent's user ID | | target_country / caller_country | Country names | | seconds_waiting_in_queue | Queue wait time in seconds (queue calls only) | | OriginalIvrUniqueID | Original call ID (for transferred calls) | | DepartmentName / DepartmentID | Account/department info | | IVR | Array of IVR layers the call traversed (layer ID, name, DTMF pressed) | | aiData | Full AI analysis (see below — only if AI is enabled on your account) |

Call Types

| Type Name | Description | |---|---| | Incoming Call | Inbound call that ended in IVR or extension (not a queue) | | Queue | Inbound call received by a queue | | Extension Outgoing | Manual outgoing call from extension | | Click2Call leg1 | Click2Call — Leg 1 (connection to agent) | | Click2Call leg2 | Click2Call — Leg 2 (connection to customer, has recording) | | ProductiveCall Leg1/Leg2 | Agent Auto Dialer calls | | Click 2 IVR / Click 2 IVR Incoming | Predictive Dialer calls | | XferCDR | Manually transferred call | | VoiceMail | Call answered by voicemail |

Call Statuses

ANSWER, NOANSWER, BUSY, CANCEL, ABANDONE, TIMEOUT, FULL, EXIT, JOINEMPTY, VOEND, TE, NOTCALLED, VOICEMAIL

Full JSON Example (Outgoing Call)

{
  "caller": "0722776772",
  "target": "0501234567",
  "time": 1595960350,
  "duration": 11,
  "ivruniqueid": "2020072818dcDHFJcc804",
  "type": "Extension Outgoing",
  "status": "ANSWER",
  "callerextension": "SIPSIP",
  "did": "",
  "queueid": 0,
  "queuename": "",
  "record": "https://cpanel.voicenter.co.il/CallsHistory/PlayRecord/2020072818dcDHFJcc804-aws.mp3",
  "price": 0,
  "dialtime": 2,
  "representative_name": "John Doe",
  "representative_code": "12345678",
  "DepartmentID": 12345678,
  "DepartmentName": "Voicenter Sales"
}

Full JSON Example (Incoming Queue Call)

{
  "caller": "0501234567",
  "target": "",
  "time": 1595333610,
  "duration": 20,
  "ivruniqueid": "202007211213270124c",
  "type": "Queue",
  "status": "ANSWER",
  "did": "0722776772",
  "queueid": 123456789,
  "queuename": "Sales Queue",
  "record": "https://cpanel.voicenter.co.il/CallsHistory/PlayRecord/202007211213270124c-aws.mp3",
  "seconds_waiting_in_queue": 20,
  "IVR": [
    { "layer_id": 1234, "layer_name": "Main IVR", "layer_number": 0, "Dtmf": 2, "dtmf_order": 1 },
    { "layer_id": 4321, "layer_name": "Sales Department", "layer_number": 2, "Dtmf": 0, "dtmf_order": 2 }
  ]
}

AI Data (if enabled)

When AI analysis is enabled on the account, aiData contains:

{
  "aiData": {
    "insights": {
      "questions": [
        { "key": "category", "question": "Categorize the call...", "data_type": "string", "answer": "technical issue" },
        { "key": "issue_resolved", "question": "Was the issue resolved?", "data_type": "boolean", "answer": "false" }
      ],
      "participants": {
        "caller": { "name": null, "personality_traits": [] },
        "callee": { "name": "Steve", "personality_traits": ["professional", "patient"] }
      },
      "summary": "The caller reported dropped calls. The agent ran through troubleshooting steps..."
    },
    "emotions": {
      "sentences": [
        { "sentence_id": 21, "emotion": "frustrated", "emotion_direction": -1, "confidence_emotion": 0.92, "intensity_emotion": 0.85 }
      ]
    },
    "transcript": [
      { "speaker": "Speaker0", "text": "Good afternoon, thank you for calling support", "startTime": 1.87, "endTime": 5.2, "sentence_id": 0 },
      { "speaker": "Speaker1", "text": "Hi, I'm having issues with my phone service", "startTime": 13.46, "endTime": 16.6, "sentence_id": 1 }
    ]
  }
}

• Speaker0 = agent (callee), Speaker1 = customer (caller)
• emotion_direction: 1 = positive, -1 = negative, 0 = neutral
• data_type values: 1=boolean, 2=string, 3=number, 4=json array, 5=json object list, 6=json object

TypeScript Implementation (Express)

import express, { Request, Response } from 'express';

const app = express();
app.use(express.json());

interface CdrPayload {
  ivruniqueid: string;
  caller: string;
  target: string;
  type: string;
  status: string;
  duration: number;
  actualCallDuration: number;
  record: string;
  representative_name: string;
  representative_code: string;
  did: string;
  queuename: string;
  seconds_waiting_in_queue?: number;
  IVR?: Array;
  aiData?: {
    insights?: { summary: string; questions: Array };
    transcript?: Array;
    emotions?: { sentences: Array };
  };
}

app.post('/webhooks/voicenter/cdr', async (req: Request, res: Response) => {
  const cdr = req.body as CdrPayload;

  // 1. Acknowledge immediately — Voicenter needs a fast response
  res.json({ Err: 0, Errdesc: 'OK' });

  // 2. Process asynchronously after responding
  setImmediate(async () => {
    try {
      // Deduplicate — Voicenter may retry on network errors
      const exists = await db.calls.findOne({ callId: cdr.ivruniqueid });
      if (exists) return;

      // Save CDR to database
      await db.calls.create({ callId: cdr.ivruniqueid, ...cdr });

      // Missed queue call → trigger SMS callback offer
      if (cdr.status === 'ABANDONE' && cdr.queuename) {
        await sendSms(cdr.caller, 'We missed your call. We will call you back shortly.');
      }

      // Save AI analysis if available
      if (cdr.aiData?.insights?.summary) {
        await db.callInsights.create({
          callId: cdr.ivruniqueid,
          summary: cdr.aiData.insights.summary,
          questions: cdr.aiData.insights.questions,
          transcript: cdr.aiData.transcript,
        });
      }
    } catch (err) {
      console.error('CDR processing failed:', err);
    }
  });
});

app.listen(3000);

Your Required Response

{ "Err": 0, "Errdesc": "OK" }

| Err | Meaning | |---|---| | 0 | OK | | 1 | Parse error | | 2 | Application error |

Tips

• Respond immediately before processing — Voicenter waits for your 200 OK. Use setImmediate, a queue job, or background worker for heavy processing.
• Idempotency — Voicenter may retry on network failures. Store ivruniqueid as a unique key and skip duplicates.
• Recording URL — The record field is a direct MP3 link. Store it in your CRM for playback.
• Queue abandons — status: ABANDONE + queuename set = customer hung up while waiting. Send them an SMS/WhatsApp callback offer.

Related Skills

• Call Log API — Query historical CDRs on-demand instead of receiving them via webhook
• Pop-Up Screen — Uses ivruniqueid from CDR to correlate with the ringing-phase popup
• Real-Time API — Receive call events live during the call (before it ends)
• Blacklist — Automatically blacklist numbers that opt out via DTMF during a call

Source & license

This open-source skill is cataloged on AgentStack and links to its original source — we do not rehost the code.

• Author: VoicenterTeam
• Source: VoicenterTeam/claude-marketplace
• License: MIT

Install and usage instructions live in the source repository linked above.