custovox-calls
An MCP server that lets Claude analyze customer-service calls: sentiment, contact reasons, complaints, churn threats and agent promises, then suggest follow-up actions. Everything runs on your machine.
Built on callscope-ai.
Install
pip install custovox-calls # transcripts (JSON segments) and demo mode
pip install "custovox-calls[whisper]" # + local transcription of audio files (faster-whisper)
Use in Claude Code
# try it with 6 fictional calls, no audio needed
claude mcp add custovox-calls-demo --env CUSTOVOX_CALLS_DEMO=1 -- uvx custovox-calls
# your own calls
claude mcp add custovox-calls -- uvx --from "custovox-calls[whisper]" custovox-calls
Then ask: "Analyze ~/calls/2026-11-18-0912.wav", "What did customers complain about this month?", "Which calls had a churn threat?", "What should we do after call demo-1?"
Use in Claude Desktop
{
"mcpServers": {
"custovox-calls": {
"command": "uvx",
"args": ["--from", "custovox-calls[whisper]", "custovox-calls"]
}
}
}
Tools
| Tool | What it does |
|---|---|
analyze_call(source, call_id, started_at, language) |
Analyze one call and save it. source: local audio file, http(s) URL to an audio file, local JSON file of segments [{start, end, text}], or a demo call (demo-1 … demo-6) |
get_call_insights(since, until) |
Summary over a period: calls, minutes, mean customer sentiment, contact reasons, churn threats, most negative calls |
search_feedback(query, speaker, since, until) |
Find what customers (or agents) said; case and accents ignored |
suggest_actions(call_id) |
Manager alert, customer callback, CRM ticket, each with its reason and the sentence that triggered it. Sends nothing |
send_action(action_id) |
Only when CUSTOVOX_CALLS_WEBHOOK_URL is set: posts one suggested action to your webhook (Slack, Zapier, Make, a CRM…) |
Configuration
| Variable | Default | Meaning |
|---|---|---|
CUSTOVOX_CALLS_DB |
~/.custovox-calls/calls.db |
SQLite file holding the analyzed calls |
CUSTOVOX_CALLS_DEMO |
off | 1: in-memory database preloaded with 6 fictional calls |
CUSTOVOX_CALLS_WEBHOOK_URL |
none | Enables send_action. The JSON payload has a text field, so Slack incoming webhooks display it as is |
CUSTOVOX_CALLS_WHISPER_MODEL |
small |
faster-whisper model size |
Privacy and limits
- Call recordings and transcripts contain personal data. Audio is transcribed locally and the analysis is stored in a local SQLite file; nothing is sent anywhere, except by
send_actionto the webhook you configured. - Speaker detection (agent or customer), sentiment and contact reasons come from simple, inspectable rules in callscope-ai, in French and English. Use them to sort and prioritize calls, and check important cases by listening to them.
Development
pip install -e ".[test]" && pytest # no network, no audio, no model
MIT licensed.
Need more?
custovox.ai — Full customer intelligence platform. From $89/month. https://custovox.ai
Metadata
Release files for custovox-calls 0.1.0
For a detailed explanation of source distributions (sdists) and built distributions (wheels), please see the package formats documentation.
Source distribution (sdist)
| File | Size | Uploaded | Provenance | Actions |
|---|---|---|---|---|
| custovox_calls-0.1.0.tar.gz | 13.6 kB | Without provenance | Details for custovox_calls-0.1.0.tar.gz |
Built distribution (wheel)
| File | Interpreter | ABI | Platform | Actions Reset |
|---|---|---|---|---|
| custovox_calls-0.1.0-py3-none-any.whl | Python 3 | none | any | Details for custovox_calls-0.1.0-py3-none-any.whl |
Total release size: 26.1 kB
Release files / custovox_calls-0.1.0.tar.gz
| Download URL | custovox_calls-0.1.0.tar.gz |
|---|---|
| Size | 13.6 kB |
| Tags | Source |
|
SHA-256 checksum How to use checksums |
bf758e85db0d7553a1c34a1fc68909c4424033d19d5bbafac6f0244d6db49213
|
|
BLAKE2b-256 checksum How to use checksums |
3a2ad613b4ebcd261dc25b4fdc84c85076b9f5f34093945623c1ad2a133d5bcd
|
| Upload date | |
|
Uploaded using Trusted Publishing? What is trusted publishing? |
No |
| Uploaded via |
twine/7.0.0 CPython/3.12.10
|
Release files / custovox_calls-0.1.0-py3-none-any.whl
| Download URL | custovox_calls-0.1.0-py3-none-any.whl |
|---|---|
| Size | 12.5 kB |
| Tags | Python 3 |
|
SHA-256 checksum How to use checksums |
9f2068fc1e6cdd13798440ace64c825eab996d846bc1a1643815c9801b161038
|
|
BLAKE2b-256 checksum How to use checksums |
687b2770a1fa09574b3a8b39b12810f64684b7e7d9b9572811d6d6596d75776b
|
| Upload date | |
|
Uploaded using Trusted Publishing? What is trusted publishing? |
No |
| Uploaded via |
twine/7.0.0 CPython/3.12.10
|
Security for custovox-calls 0.1.0
Provenance
Provenance describes where a file came from. On PyPI, provenance is shared via attestations, which provide a verifiable record of the build or publishing details. View details, limitations and caveats.
No provenance found
None of the files in this release were uploaded with provenance metadata.
A note on package security
PyPI is a repository for distributing packages, not a service for auditing them. While third-party security researchers and our staff do their best to detect and remove malicious packages, PyPI cannot guarantee the quality, security or suitability of any given package.
To learn more about protecting yourself from malicious actors, visit our package security documentation.