What problem does it solve? Setting up a voice agent testing or observability workflow on the Cekura platform involves many steps — connecting a provider, configuring metrics, generating evaluators, and running a first call — and it is easy to end up with records created but nothing actually verified. This Skill walks a new user through a structured, phase-by-phase onboarding that only ends at a verified working loop: a completed test call with a visible transcript, or an ingested and scored production call. ## Core Features & Use Cases - Two onboarding paths: a testing path (generate evaluators and run simulated calls against the agent) and an observability path (ingest production call logs and evaluate them with metrics). - Provider-aware agent connection: supports VAPI, Retell, ElevenLabs, Synthflow, LiveKit, Pipecat, Bland, KoreAI, Genesys, Cisco, and self-hosted stacks, with auto-import of agent configuration where available and a credential matrix per provider. - Verification gates at every phase: each step has an explicit gate condition (agent created, metrics enabled, evaluators generated, call connected with a two-sided transcript) so misconfigurations surface during onboarding rather than later. - Use Case: A team that just built a voice agent on VAPI signs up for Cekura. The Skill auto-imports the agent from their VAPI API key, confirms default metrics, generates ten behavioral evaluators from the agent's system prompt, runs one test call, and shows the transcript and metric scores — all in one guided session. ## Quick Start Ask your AI assistant to help you get started with Cekura and walk you through connecting your voice agent and running your first test call.
