skills.rest
cekura-ai
cekura-aiOFFICIAL
Explore all skills by this author →
WebsiteGitHub

cekura-infra-test-suite

Creates source-controlled Cekura JSON test suites and CI gates for voice agent pipelines.

7|2|Updated Mar 6, 2026
Search Tags:#ci/cd#test automation#github actions#regression testing#voice ai#cekura#json spec
One-click install
npx skills add https://github.com/cekura-ai/cekura-skills --skill cekura-infra-test-suite-cekura-ai
Or copy as Structured Prompt for Agent▼
Please help me install this Agent Skill.
Skill: cekura-infra-test-suite
Source: https://github.com/cekura-ai/cekura-skills/tree/main/cekura/skills/cekura-infra-test-suite
Command: npx skills add https://github.com/cekura-ai/cekura-skills --skill cekura-infra-test-suite-cekura-ai

SYSTEM DOCUMENTATION & REQUIREMENTS

💡 This Skill includes scripts (resource) and references (resource) components.

What problem does it solve? Voice AI teams lack deterministic regression coverage for their STT, LLM, TTS, VAD, interruption, and call-lifecycle pipelines, so code changes silently break call behavior. This Skill inspects a voice-agent repository, authors a compact Tests-as-Code JSON suite, validates it against Cekura, and wires a CI workflow that blocks merges when the voice pipeline regresses. ## Core Features & Use Cases - Repository-driven test authoring: Traces the runtime call path (transport, STT, turn detection, idle timers, DTMF, tools) and turns sourced thresholds and quoted agent phrases into transcript-verifiable assertions. - Offline linting and safe validation: Runs scripts/lint_suite.py for free offline checks, then validates the spec with Cekura's scenarios_validate_json endpoint without placing calls or spending credit. - CI gate wiring: Generates a GitHub Actions (or GitLab/Jenkins) workflow with manual dispatch, a dry-run checkbox, and polling logic that fails the build on any failed call. - Use Case: After modifying your Pipecat bot's interruption handling, ask the Skill to update cekura.tests.json so CI proves the barge-in behavior still works before merge. ## Quick Start Ask the assistant to inspect this voice agent repository and create a Cekura CI test suite covering its interruption, idle-timer, and happy-path behavior.

Frequently Asked Questions about cekura-infra-test-suite

High-intent search queries and answers about installing and using this skill.

FAQPage Schema
How do I add voice agent regression tests to CI?▼

Run this Skill against your voice agent repository. It inspects the call pipeline, authors a cekura.tests.json spec with deterministic conditional-actions cases, validates it against Cekura, and writes a GitHub Actions workflow that polls runs and fails the build on regressions.

How do I test interruption and barge-in behavior in a voice bot?▼

Use conditional_actions cases with the self-closing <interruption time="Xs" /> tag at the start of an action_followup condition. The Skill's case catalog includes an interruption gauntlet covering backchannels, decisive interrupts, and barges after silence.

Does the Cekura test suite work with GitLab or Jenkins instead of GitHub Actions?▼

Yes. The default deliverable is a GitHub Actions workflow, but the Skill substitutes a .gitlab-ci.yml, Jenkinsfile stage, or Makefile target when the repository uses a different CI system, keeping the same validate-then-poll structure.

Can I validate a Cekura test spec without placing real calls?▼

Yes. Run scripts/lint_suite.py offline for free structural and authoring-rule checks, then use the scenarios_validate_json endpoint or run_suite.py --dry-run, which validates metrics, personalities, and channel compatibility without dialing or spending credit.

Why does my Cekura spec validation fail on metrics?▼

A spec can only reference metrics that exist and are enabled for the target agent. The Skill reads the agent's enabled metrics before authoring; an unenabled or invented slug is rejected at validation and must be fixed in the workspace, not removed from the case.

When should I use cekura-eval-design instead of this Skill?▼

Use cekura-eval-design for one-off dashboard evaluators or adaptive quality scenarios. This Skill is for source-controlled Tests-as-Code suites that live in the voice-agent repository and gate CI on pipeline regressions.

Related Skills

Explore More Software Engineering Skills→
pipecat-aipipecat-ai
132

pipecat

Start Pipecat MCP server voice sessions with start(), speak(), listen(), and stop().

Official
Intermediate
svangesvange
1

ai-monitor-pipeline

Monitor CI pipelines and triage failures after code pushes.

Community
Intermediate
cekura-aicekura-ai
5

Cekura Eval Design

Design, configure, and execute evaluators for AI voice agents on Cekura.

Official
Advanced
BaryoDevBaryoDev

mcp-builder

Create MCP servers in Python or TypeScript with schema validation.

Official
Advanced