cekura-metric-design

Use when the user asks to "create a metric", "write a metric", "design a metric", "build a metric for", "evaluate agent performance", "measure call quality", "track a KPI", "add a workflow metric", "improve my metric", "fix a metric", "debug metric results", "set up quality scoring", or "what metrics do I need". Also relevant when discussing LLM judge prompts, custom code metrics, evaluation triggers, VALID_SKIP patterns, section extraction, or metric best practices for Cekura voice AI agents. Covers both creating new metrics and reviewing, iterating on, or troubleshooting existing ones.

1.38x

Quality

—

Does it follow best practices?

Impact

98%

1.38x

Average score across 3 eval scenarios

Securityby

Advisory

Suggest reviewing before use

Quality

Content

62%

Reviews the quality of instructions and guidance provided to agents. Good implementation is clear, handles edge cases, and produces reliable results.

The body is a well-structured, mostly lean guide with a clear creation workflow and explicit validation checkpoints for the cost-sensitive evaluation context. Its main weaknesses are scaffolded (placeholder) prompt templates whose executable counterparts live in referenced example files that are missing from the bundle, and some explanatory prose that could be trimmed.

Suggestions

Add the four referenced example files under examples/ (llm-judge-metric.md, narrative-metric.md, custom-code-metric.py, section-extraction-metric.py) — currently these references point to files that do not exist, breaking progressive disclosure.

Tighten explanatory asides (e.g., the 'Expected Outcome is transcript-only' paragraph) to bare-minimum notes so the body earns its token budget, or move the detail into a reference file.

Inline at least one complete, copy-paste-ready metric prompt example in the body so the core actionability does not depend solely on the missing example files.

Dimension	Reasoning	Score
Conciseness	The body is largely lean and assumes Claude's competence (no basic-concept padding), but at ~240 lines it includes explanatory asides that could be tightened (e.g., the multi-sentence 'Expected Outcome is transcript-only' note), so it lands at 'mostly efficient but could be tightened' rather than the lean top anchor.	2 / 3
Actionability	Concrete specifics are present (real template variables, exact API params like page_size=1/up to 200 and timestamp filters, PASS/FAIL examples), but the prompt templates are scaffolds with placeholders ([Positive indicator 1]) and the fully executable metric examples are pointed at example files rather than being inline; this matches 'some concrete guidance but incomplete' rather than copy-paste-ready.	2 / 3
Workflow Clarity	The Metric Creation Workflow is a clear numbered 6-step sequence with explicit feedback checkpoints ('Deploy and test', 'Iterate... Plan for at least one iteration'), and the Manual-Fix-First flow adds re-evaluation on 20-30 calls; the cost-sensitive batch context has an explicit pre-evaluation validation step, matching the top anchor.	3 / 3
Progressive Disclosure	The overview is well-structured with clearly signaled one-level-deep references into references/ (all four files exist), but four example files under examples/ are referenced and do not exist in the bundle, so navigation is partially broken — 'some structure but could be better organized' rather than the fully clean top anchor.	2 / 3
	Total	9 / 12 Passed

Description

100%

Based on the skill's description, can an agent find and select it at the right time? Clear, specific descriptions lead to better discovery.

The description is specific, trigger-rich, and complete, explicitly stating both what the skill covers and when to invoke it with many natural user phrasings. It is well-distinctified within the Cekura skill family. Voice is appropriately third person ('Use when the user asks'), so no specificity penalty applies.

Dimension	Reasoning	Score
Specificity	Lists multiple specific concrete actions ('create a metric', 'evaluate agent performance', 'measure call quality', 'track a KPI', 'LLM judge prompts, custom code metrics, evaluation triggers, VALID_SKIP patterns, section extraction'), matching the top anchor rather than the partial 'names domain and some actions' anchor below.	3 / 3
Completeness	Explicitly answers both 'what' ('Covers both creating new metrics and reviewing, iterating on, or troubleshooting existing ones') and 'when' via a thorough 'Use when the user asks to...' clause, matching the top anchor rather than the score-2 case where 'when' is only implied.	3 / 3
Trigger Term Quality	Covers natural phrases users would actually say ('fix a metric', 'improve my metric', 'what metrics do I need', 'measure call quality', 'track a KPI') plus technical variations, matching the 'good coverage of natural terms' anchor; not the level below which only has some keywords missing common variations.	3 / 3
Distinctiveness Conflict Risk	Niche is clearly scoped to 'Cekura voice AI agents' metrics with distinct trigger phrases and cross-references to sibling cekura-eval-design/improvement skills, so it is unlikely to fire for an unrelated skill; not the score-2 anchor where overlap with similar skills remains.	3 / 3
	Total	12 / 12 Passed

Validation

100%

Warnings & errors only

Checks the skill against the spec for correct structure and formatting. All validation checks must pass before discovery and implementation can be scored.

Validation — 16 / 16 Passed

Validation for skill structure

No warnings or errors.

Repository: cekura-ai/cekura-skills
Commit: f0854af

Reviewed: about 24 hours ago

Table of Contents

Discovery Implementation Validation

Is this your skill?

If you maintain this skill, you can claim it as your own. Once claimed, you can manage eval scenarios, bundle related skills, attach documentation or rules, and ensure cross-agent compatibility.