E2E Testing for
LiveKit / Pipecat Voice Agents
You built a custom voice agent on LiveKit or Pipecat. Now add the testing layer it deserves. vspec calls your SIP endpoint or PSTN number and validates every conversation scenario.
Custom stack.
Platform-agnostic testing.
Point at your phone number
Whether your LiveKit or Pipecat agent answers via SIP trunking or a PSTN number, vspec calls it directly. No changes to your infrastructure required.
Write scenario-based tests
Define what a caller says, what your agent must do or say in response, and what outcomes constitute a pass. As complex as your pipeline, as simple as a YAML file.
Integrate with your deploy pipeline
Trigger vspec from GitHub Actions, GitLab CI, or any webhook. Get a structured test report before code reaches production.
Built on open source.
Tested like production.
Pipeline regression detection
Pipecat pipelines chain STT, LLM, and TTS together. Any component update can break the full flow. vspec catches this before users do.
LLM prompt integrity
Test that your LiveKit agent's system prompt produces the right behaviors in a real call — not just in a chat playground.
Tool use & function calling
Verify that your agent's tool calls execute correctly mid-conversation — database lookups, API calls, memory retrieval — under real latency conditions.
Latency-aware testing
Real calls expose timing issues that unit tests miss. vspec tests run over actual telephony infrastructure, surfacing latency-related failures early.
Your custom stack deserves
a proper test suite.
Free tier. No credit card. First test in under 2 minutes.
Start testing for free →