llm-evaluation

Implement comprehensive evaluation strategies for LLM applications using automated metrics, human feedback, and benchmarking. Use when testing LLM performance, measuring AI application quality, or establishing evaluation frameworks.

Quality

75%

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/llm-evaluation/SKILL.md

Quality

Content

50%Weight 40%Scale 1-3

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 broad, code-heavy catalog of evaluation techniques that is largely actionable, but it is weakened by a fictional Quick Start library, a few broken/incomplete snippets, generic boilerplate sections, no sequenced validation workflow, and references to bundle files that do not exist.

Suggestions

Replace the fictional `from llm_eval import EvaluationSuite, Metric` Quick Start with a real, runnable example using the actual libraries shown later (nltk/rouue_score/bert_score), and implement or remove the `pass`-only `calculate_factuality`.

Add a sequenced evaluation workflow with explicit validation checkpoints (define metrics → run on held-out data → check for regressions/statistical significance → fix and re-run) instead of presenting metrics as an unstructured catalog.

Either create the referenced bundle files (references/metrics.md, assets/evaluation-framework.py, scripts/evaluate-model.py, etc.) or remove the Resources entries, and move the detailed per-metric implementations into those files so SKILL.md stays a lean overview.

Trim the generic boilerplate (the "Do not use this skill when" and "Instructions" sections, the opening restatement of the description) to improve token efficiency.

Dimension	Reasoning	Score
Conciseness	Mostly efficient code, but padded with boilerplate ("Do not use this skill when" generic bullets, generic "Instructions" list, an opening line that restates the description) that adds little skill-specific value.	2 / 3
Actionability	Much of the code (nltk BLEU, rouge_score, bert_score, scipy, sklearn) is executable, but the Quick Start imports a fictional "llm_eval" package, calculate_factuality is a bare "pass", and the inter-rater code has a broken "interpretation[True]" lookup.	2 / 3
Workflow Clarity	Content is organized by topic but lacks a sequenced end-to-end evaluation workflow with validation checkpoints; destructive/batch-style operations (regression detection, A/B testing) have no validate-then-fix feedback loops.	2 / 3
Progressive Disclosure	A Resources section signals one-level-deep references, but all nine referenced bundle files (references/, assets/, scripts/*) are missing, and detailed metric implementations are inlined in SKILL.md rather than split out.	2 / 3
	Total	8 / 12 Passed

Description

100%Weight 40%Scale 1-3

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 strong: it states concrete capabilities, includes an explicit "Use when..." trigger clause with natural keywords, and carves out a distinct niche. Minor "comprehensive" buzzword aside, it cleanly satisfies all dimensions.

Dimension	Reasoning	Score
Specificity	Names multiple concrete evaluation approaches — "automated metrics, human feedback, and benchmarking" — rather than vague language, matching the anchor for listing several specific actions.	3 / 3
Completeness	Explicitly answers both what ("Implement comprehensive evaluation strategies...") and when ("Use when testing LLM performance..."), with an explicit trigger clause.	3 / 3
Trigger Term Quality	"Use when testing LLM performance, measuring AI application quality, or establishing evaluation frameworks" gives good coverage of natural terms a user would say when needing this skill.	3 / 3
Distinctiveness Conflict Risk	Occupies a clear LLM-evaluation niche with distinct triggers unlikely to fire for unrelated skills.	3 / 3
	Total	12 / 12 Passed

Validation

93%

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 — 15 / 16 Passed

Validation for skill structure

Criteria	Description	Result
referenced_paths_exist	Referenced path issues: 8 missing	Warning

	Total	15 / 16 Passed

Repository: rmyndharis/antigravity-skills
Path: skills/llm-evaluation/SKILL.md
Commit: e63f7dd

Reviewed: 13 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.