CtrlK
BlogDocsLog inGet started
Tessl Logo
SkillPublic

second-opinion

Gets a blind review of a diff or a PR from a model of another provider (Claude, GPT, Gemini), then checks and compares its findings. Use when the user asks for a second opinion, an adversarial review, or a review by another model.

Quality

88%

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

SKILL.md
Quality
Evals
Security

Quality

Content

92%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.

A lean, highly actionable workflow: six sequenced steps with per-step done-criteria, a built-in verification loop for the reviewer's findings, concrete CLI commands, and verbatim brief language. The only gap is that the delegation-tool launch path is left abstract and no example brief is included.

DimensionReasoningScore

Conciseness

The 28-line body is terse and imperative throughout ('Write your own view first... Keep it out of the brief'), with no explanations of concepts Claude already knows. The three-sentence intro defines the blind-review constraint that governs the whole procedure, so every token earns its place — anchor 5, not 4, since nothing needs trimming.

5 / 5

Actionability

Provides concrete, near copy-paste commands ('codex exec -s read-only -m <model> -o <reply-file> - < <brief-file>', 'claude -p --permission-mode plan --model <model>'), verbatim scope and output strings, and a bulleted brief structure. Anchor 4 rather than 5: the first launch option ('A delegation tool of your agent environment') is abstract and no worked example brief is given; well above the pseudocode level of 3.

4 / 5

Workflow Clarity

Six clearly sequenced steps, each with an explicit completion criterion ('Done when you can name it by ref or URL', 'Done when every finding has a verdict with a reason'), a verification loop in step 6 (confirmed/disputed with a reason, plus a rule for findings that name no producer), and fallbacks when no other provider is available (steps 2 and 5). The operation is read-only, so the destructive/batch cap does not apply. Anchor 5, not 4, because checkpoints are explicit rather than mostly present.

5 / 5

Progressive Disclosure

The skill is under 50 lines with well-organized sections and no bundle files, and its sole reference — 'the section "Edge cases and guards" from the [code-review rules](../code-review/rules.md)' — is clearly signaled, names an exact section, and is one level deep (it points to a sibling skill in the same collection, not a nested bundle file). Qualifies for 5 under the simple-skill exception.

5 / 5

Total

19

/

20

Passed

Description

83%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.

A strong description that clearly and explicitly states both the capability (blind cross-provider review of a diff or PR, followed by checking and comparing findings) and concrete usage triggers. The only weaknesses are minor: a few natural trigger synonyms are absent and 'adversarial review' slightly overlaps with generic review skills.

DimensionReasoningScore

Specificity

Names the domain (diff/PR review by 'a model of another provider (Claude, GPT, Gemini)') and three concrete actions: 'Gets a blind review', 'checks', and 'compares its findings'. Not 3, since more than 1-2 actions are explicitly stated; not 5, since supporting actions like writing the brief or reporting the comparison are not covered.

4 / 5

Completeness

Explicitly answers both: what ('Gets a blind review of a diff or a PR from a model of another provider... then checks and compares its findings') and when ('Use when the user asks for a second opinion, an adversarial review, or a review by another model') with concrete trigger phrases — a direct match to the anchor-5 example pattern. Not 4, because the 'when' clause is already explicit and specific.

5 / 5

Trigger Term Quality

The trigger phrases 'second opinion', 'an adversarial review', and 'a review by another model' are natural terms a user would say. Good coverage, but a few natural variants are missing (e.g. 'another set of eyes', 'have GPT/Gemini look at this'), so it falls between anchors 4 and 5, closer to 4.

4 / 5

Distinctiveness Conflict Risk

Clear niche (cross-provider blind review) with distinct triggers like 'second opinion' and 'review by another model', but 'an adversarial review' could plausibly fire a generic code-review skill — minor overlap with a closely related skill, matching anchor 4 rather than 5.

4 / 5

Total

17

/

20

Passed

Validation

81%

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

Validation — 13 / 16 Passed

Validation for skill structure

CriteriaDescriptionResult

metadata_version

'metadata.version' is missing

Warning

metadata_field

'metadata' should map string keys to string values

Warning

relative_links

Relative link issues: 1 suspicious

Warning

Total

13

/

16

Passed

Repository

storybookjs/storybook

Path

.agents/skills/second-opinion/SKILL.md

Reviewed

Table of Contents

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.