evaluation

This skill should be used when building agent evaluation systems: deterministic checks, regression suites, multi-dimensional rubrics, quality gates, production monitoring, baseline comparison, and outcome measurement for agent pipelines.

Quality

71%

Does it follow best practices?

Run evals on this skill

Adds up to 20 points to the overall score

View guide

Securityby

Passed

No findings from the security scan

Fix and improve this skill with Tessl

tessl review fix ./skills/evaluation/SKILL.md

Quality

Content

67%Weight 40%Scale 1-5

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

The body is well-structured, actionable, and includes validation-aware workflows with concrete thresholds and code. Its main weakness is thematic redundancy across sections that hurts conciseness, plus an unreferenced bundle script and illustrative-only code examples.

Suggestions

Consolidate the repeated advice about multi-dimensional rubrics, outcomes-over-paths, and complexity stratification that currently appears in Core Concepts, Guidelines, and Gotchas into a single authoritative section.

Reference scripts/evaluator.py from the body (e.g. in Examples or Practical Guidance) so the bundled evaluator is discoverable, or remove it from the bundle.

Tighten code examples toward executability by defining or stubbing helpers like load_rubric() and assess_dimension(), or label them explicitly as illustrative pseudocode.

Dimension	Reasoning	Score
Conciseness	The body assumes Claude's intelligence and avoids basic-concept padding, but the same themes (multi-dimensional rubrics, outcomes over paths, complexity stratification) are restated across Core Concepts, Detailed Topics, Guidelines, and Gotchas, producing more than minor redundancy.	3 / 5
Actionability	Provides concrete code examples, specific thresholds (0.7/0.9/0.85/0.70), named dimensions with weights, and a 9-step framework, but the code snippets rely on undefined helpers like load_rubric() and assess_dimension(), making them illustrative rather than copy-paste complete.	4 / 5
Workflow Clarity	The 'Building Evaluation Frameworks' section gives a clearly sequenced 9-step build flow and the deterministic-gate example shows a validate-then-judge feedback loop, but error-recovery branches live in examples rather than inline in the main workflow, leaving minor validation gaps.	4 / 5
Progressive Disclosure	Content is well-sectioned and the one body reference (./references/metrics.md) is one-level-deep with a clear 'Read when' signal, but scripts/evaluator.py exists as a bundle file yet is never referenced from the body, a minor navigation gap.	4 / 5
	Total	15 / 20 Passed

Description

75%Weight 40%Scale 1-5

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 and comprehensive, naming seven concrete evaluation activities and covering both what and when. Its main weaknesses are a tautological 'when' clause, missing common synonyms like 'benchmarks'/'evals', and a directive voice that keeps most dimensions at 4 rather than 5.

Suggestions

Replace the tautological 'when building agent evaluation systems' with concrete user-side triggers, e.g. 'Use when the user asks to set up evals, build regression suites, or add quality gates to an agent pipeline.'

Add natural synonyms users actually say ('evals', 'benchmarks', 'testing') to broaden trigger-term coverage.

Rephrase in third person ('Builds and runs agent evaluation systems...') to avoid the second-person/directive voice penalty.

Dimension	Reasoning	Score
Specificity	Lists seven concrete actions ('deterministic checks, regression suites, multi-dimensional rubrics, quality gates, production monitoring, baseline comparison, and outcome measurement') giving comprehensive coverage, but the directive 'This skill should be used' voice trims it from a clean 5 per the second-person penalty.	4 / 5
Completeness	Explicitly states both what ('building agent evaluation systems: ...') and when ('should be used when building agent evaluation systems'), but the 'when' is tautological rather than concrete user-side trigger phrases, so it does not reach the explicit-trigger level 5.	4 / 5
Trigger Term Quality	Strong natural evaluation vocabulary ('regression suites', 'quality gates', 'production monitoring', 'baseline comparison') but missing common synonyms users say such as 'testing', 'benchmarks', or 'evals', so it sits at good-not-comprehensive coverage.	4 / 5
Distinctiveness Conflict Risk	'Agent evaluation systems' with outcome-measurement framing is a clear niche, but 'evaluation' broadly overlaps with the adjacent advanced-evaluation skill, leaving minor conflict risk rather than minimal.	4 / 5
	Total	16 / 20 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: muratcankoylan/Agent-Skills-for-Context-Engineering
Path: skills/evaluation/SKILL.md
Commit: 6dbe1a1

Reviewed: 19 days 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.