CtrlK
BlogDocsLog inGet started
Tessl Logo

testland/openai-evals

Authors and runs OpenAI Evals - Python framework + registry for evaluating LLMs and LLM-backed systems with `oaieval {model} {eval-name}` CLI; supports template-based evals (Match / Includes / FuzzyMatch / ModelBasedClassify) defined in `evals/registry/evals/*.yaml` against JSONL data files in `evals/registry/data/`, plus custom Python eval classes implementing the Eval interface. Use when the user works with the openai/evals repo, needs the OpenAI-curated eval registry, or contributes new evals via PR to the registry.

80

Quality

100%

Does it follow best practices?

Run evals on this skill

Adds up to 20 points to the overall score

View guide

SecuritybySnyk

Low

Low-risk findings worth noting

Overview
Quality
Evals
Security
Files

Quality

Content

100%

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

A well-organized, concise, and actionable body: executable commands and paths, a sequenced install-to-CI workflow, a useful anti-patterns table, and clean one-level-deep references to upstream docs. No bundle files are present, but the in-file structure handles disclosure well.

DimensionReasoningScore

Conciseness

Lean and dense — concrete commands, file paths, a compact anti-patterns table, and a limitations list, with no padding or explanation of concepts Claude already knows. Not below 3 because every section earns its tokens; not applicable above the scale max.

3 / 3

Actionability

Fully executable guidance — `pip install evals`, `oaieval gpt-3.5-turbo test-match`, editable-install steps, and a copy-paste `jq` log-parsing command — with specific registry/data paths. Pointers to upstream docs for custom-eval authoring are appropriate disclosure, not vagueness.

3 / 3

Workflow Clarity

A clearly numbered Step 1–6 sequence from install through CI integration, with a pass-rate gate in Step 6. Eval running is not a destructive/batch operation, so the absence of an explicit validate-fix-retry loop does not cap the score.

3 / 3

Progressive Disclosure

No bundle files exist; the body is a well-sectioned overview that signals one-level-deep references to upstream repo docs (docs/build-eval.md, docs/custom-eval.md, docs/eval-templates.md) and alternative skills in a dedicated References section.

3 / 3

Total

12

/

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.

A specific, third-person description with concrete capabilities and an explicit 'Use when...' trigger clause scoped to the openai/evals repo. It cleanly answers both what the skill does and when to use it with minimal conflict risk.

DimensionReasoningScore

Specificity

Lists multiple concrete actions — authoring/running evals via the `oaieval` CLI, four named templates (Match/Includes/FuzzyMatch/ModelBasedClassify), and custom Python eval classes implementing the Eval interface. Uses third-person voice ('Authors and runs'), so no voice penalty applies.

3 / 3

Completeness

Explicitly answers both 'what' (authors/runs OpenAI Evals with templates and custom classes) and 'when' via a clear 'Use when the user works with the openai/evals repo...' trigger clause.

3 / 3

Trigger Term Quality

Natural user-facing terms — 'openai/evals repo', 'OpenAI-curated eval registry', 'contributes new evals via PR' — cover the phrasings a user working with this framework would actually say.

3 / 3

Distinctiveness Conflict Risk

Scoped tightly to the openai/evals repo and OpenAI-curated registry, a clear niche unlikely to trigger for unrelated evaluation skills.

3 / 3

Total

12

/

12

Passed

Validation

100%

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

Validation16 / 16 Passed

Validation for skill structure

No warnings or errors.

Reviewed

Table of Contents