A command for running Cekura evaluators against an AI agent. Evaluators are tests that assess an agent through voice, text, WebSocket, SIP, or supported voice-agent platforms such as Vapi, Retell, ElevenLabs, and LiveKit.

Use it to find evaluators by agent or project, select scenarios, determine the run mode, and execute one or more evaluations.

It turns selected evaluation scenarios into actual test runs and chooses the connection method from the agent’s configuration when needed.

run-evals — Command for Claude Code by cekura-ai

Install

Getting it into your agent

One page per mod, every tool's command on it. A separate URL per tool would split the same page into five that compete with each other.

Clone the repo

git clone --depth 1 https://github.com/cekura-ai/cekura-skills

Made for: Claude Code.

Or install cekura, the plugin that ships this one along with the rest of its 13 skills, 14 commands, 3 hooks, 1 MCP server.

Wrote this? Show the measurements

A badge with what this costs and how it scanned, read live from this page, so it follows the numbers instead of freezing them. Markdown for a README, HTML for a documentation site or a project page.

README.md

[![agentmods](https://agentmods.dev/badge/commands/cekura-ai/cekura-skills/run-evals/github.svg)](https://agentmods.dev/commands/cekura-ai/cekura-skills/run-evals)

Your own site

<a href="https://agentmods.dev/commands/cekura-ai/cekura-skills/run-evals"><img src="https://agentmods.dev/badge/commands/cekura-ai/cekura-skills/run-evals/github.svg" alt="Measured on agentmods" height="20"></a>

Or the 80×15 button, for a site that already has a row of RSS and ATOM ones. Only the verdict fits; the numbers stay here.

Your own site · 80×15

<a href="https://agentmods.dev/commands/cekura-ai/cekura-skills/run-evals"><img src="https://agentmods.dev/badge/commands/cekura-ai/cekura-skills/run-evals.svg" alt="Reviewed on agentmods" width="80" height="20"></a>

Per session 35 Only the description is in the session, so the agent can decide to use it. The body loads when it is invoked.

When invoked 2,223 The whole file, excluding the scripts and references it only reads on demand.

Security scan A 0 findings. A grade says what 26 rules found in the file — not that it is safe.

Origin original No closer match found in the catalogue.

Token cost

What it costs to keep this loaded

Counted locally with the o200k_base tokenizer, which is exact for GPT models; Claude uses its own tokenizer and its counts differ. Treat this as one consistent yardstick across the catalogue rather than a bill. Prices are per million input tokens.

Model	Per session	Once invoked
Fable 5.1	$0.00035	$0.02223
Opus 5	$0.00017	$0.01111
Sonnet 5	$0.00007	$0.00445
Haiku 4.5	$0.00003	$0.00222

Measured 3d ago against content hash 9e427cffd11e, method: parsed. Prices are Anthropic first-party input rates as of 2026-09-12, from the pricing page.

Security

Grade A, and why

run-evals scanned grade A with 0 findings against 26 rules in 11 categories — prompt injection, anti-refusal, data exfiltration, privilege escalation, supply chain, agent snooping, system-prompt leakage, SSRF and excessive agency — measured 3d ago.

A static scan of the body, not an audit. Every finding is printed with the line that produced it so you can judge whether it matters here. A mod is markdown that instructs an agent; that is exactly why what it instructs is worth reading.

Nothing flagged

None of the 26 patterns this scan looks for appear in this file: no shell pipes, no recursive deletes, no credential paths, no hidden text, no instruction-override or anti-refusal phrasing, no agent-config snooping. That is not a guarantee, it is the absence of the things that are checkable.

cekura/commands/run-evals.md · 112 lines

How it starts

The opening of the file, as written. The whole thing — 112 lines — stays where its author put it; the contents beside it link to each section on GitHub.

Tracking (do this first)

Before doing anything else, call mcp__cekura__cekura_skill_started with skill_name="run-evals". If a conversation/session ID is available (e.g. you were invoked from Cekura sandbox), also pass it as conversation_id. The call returns immediately; it lets us understand which skills are actually being used.

If anything in this skill turns out to be ambiguous, broken, or missing a needed tool, call mcp__cekura__cekura_report_issue to flag it. Use this LIBERALLY — even severity="low" reports are valuable feedback.

Run Evaluators

Execute one or more evaluators against the target agent.

Process

Identify evals to run: Get evaluator IDs or filter criteria. Use mcp__cekura__scenarios_list to find evaluators by agent or project.
Determine execution mode from agent config — don't ask if it's obvious.

If the user passed [mode] as an argument, honor it (skip detection).

Otherwise, fetch the agent with mcp__cekura__aiagents_retrieve(id=<agent_id>) and inspect provider.type, telephony.phone_number, telephony.websocket_url, provider.chat_agent_details, telephony.sip_uri. Derive candidate modes:
- voice = PSTN. Valid whenever telephony.phone_number is set. Note: a bare phone number is voice, never sip.
- sip = only when telephony.sip_uri is set (e.g. sip:agent@host).
- text = when provider.chat_agent_details is set.
- websocket = when telephony.websocket_url is set and no other provider (JSON/text protocol).
- chirp = when telephony.websocket_url is set on a voice agent (raw-PCM audio websocket).
- WebRTC (vapi, retell, elevenlabs, livekit, agora) = when provider.type matches.
- pipecat-v2 / pipecat = when provider.type: pipecat.
Selection rule:
- 0 candidates → STOP. Surface: "Agent has no provider, phone number, SIP endpoint, or websocket URL configured — can't run evals."
- 1 candidate → auto-pick. Announce: "Auto-selected <mode> — only configured connection on this agent."
- 2+ candidates → use AskUserQuestion with only the configured options, never the full list. One-line hint: text fastest/cheapest, WebRTC moderate, PSTN voice realistic but slowest.
- Pipecat exception: when the choices are pipecat and pipecat-v2, ask exactly: "Your agent uses Pipecat. pipecat (v1) uses a manually provided room URL for each evaluator run; pipecat-v2 uses configured Pipecat Cloud project credentials and creates sessions automatically." Offer only pipecat (v1) and pipecat-v2 as the options.

Read the full file on GitHub · 112 lines

Changes

What this file has done since we first saw it

Hashed on every crawl. A supply-chain change to an agent config is a question of when, not whether, so the history is kept rather than the latest state alone.

3d ago Changed · +5 lines 9e427cffd11e
12d ago First seen · 107 lines · 35 tokens per session scan A 02e6cc51f1db

Subscribe to this mod's changes

run-evals is a command published in the GitHub repository cekura-ai/cekura-skills (7 stars, last pushed yesterday), licensed MIT. It adds 35 tokens to every session and 2,223 once invoked, about $0.0002 per session on Opus 5. A static security scan graded it A with 0 findings. No closer match exists in the catalogue, so it is treated as the original; first seen 2026-08-31.

E2E testing Testing

run-evals

Getting it into your agent

Wrote this? Show the measurements

What it costs to keep this loaded

Grade A, and why

Nothing flagged

How it starts

Tracking (do this first)

Run Evaluators

Process

What this file has done since we first saw it

Other commands, from other repositories

playwright-test-heal

paul:verify

webapp-testing

ui-snapshot.template

laravel-playwright

qa