visual-ralph

Visual Ralph orchestration for frontend UI from generated references, static references, or live URL targets, using $ultragoal with built-in visual verdict and pixel-diff evidence until the implementation matches and leaves a reproducible design system.

Quality

76%

Does it follow best practices?

Run evals on this skill

Adds up to 20 points to the overall score

View guide

Securityby

Low

Low-risk findings worth noting

Fix and improve this skill with Tessl

tessl review fix ./skills/visual-ralph/SKILL.md

Quality

Content

85%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 a tight, highly actionable operational guide with an excellent gated workflow and explicit validation/feedback loops. Its main weakness is progressive disclosure: it delegates critical shared invariants to a ../../templates/AGENTS.md file that is not part of the skill's bundle, so those rules cannot be found or followed as written.

Suggestions

Ship the shared invariants inside the skill (e.g. a references/AGENTS.md one level deep) or fix the relative path so the delegated rules actually resolve within the skill's directory.

Add the concrete command (or a one-line reference to it) for running the Visual Ralph verdict and the pixel diff, mirroring the specificity already given for the imagegen continuation command.

Replace the '<command and viewport>' screenshot placeholder with a small table of common stack-specific capture commands, or an explicit instruction for deriving one from repository inspection in step 1.

Dimension	Reasoning	Score
Conciseness	Dense, operational prose with zero explanation of concepts Claude already knows; every section carries instructions (exact paths, thresholds, required JSON keys). The only mild redundancy (step 4 vs. the handoff template) is a usable fill-in artifact rather than padding, so it fits anchor 5 over anchor 4.	5 / 5
Actionability	Provides a concrete executable command ('omx imagegen continuation <session-id> --artifact ... --generated-dir "$CODEX_HOME/generated_images/<session>" --work-dir ".omx/artifacts/visual-ralph/<slug>"'), exact artifact paths, explicit thresholds ('score < 90', '>= 90'), required verdict keys, and a copy-paste handoff template. Not 5: no command is given for running the Visual Ralph verdict or pixel diff, and the screenshot command is a placeholder.	4 / 5
Workflow Clarity	Seven numbered steps in a clear sequence with an explicit approval gate (step 3), a verdict-before-every-edit checkpoint, an explicit feedback loop ('If score < 90, turn differences[] and suggestions[] into the next edit plan and rerun'), and a completion checklist with stop conditions and a concrete-blocker path. Matches the anchor-5 example including validation, error-recovery loop, and checklist.	5 / 5
Progressive Disclosure	Sections are well organized and the body correctly delegates shared invariants instead of duplicating them, but the single external reference points to '../../templates/AGENTS.md' — a path outside the skill directory that does not exist in the bundle — leaving the delegated rules unreachable from the skill. Fits anchor 3 (references present but not reliably navigable); not 4 because a broken out-of-bundle reference is more than a minor organization gap.	3 / 5
	Total	17 / 20 Passed

Description

58%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 communicates a specific, differentiated pipeline (visual reference → $ultragoal implementation → verdict/pixel-diff → design system) in third person, but has no explicit trigger guidance. Its weakest points are the missing 'Use when' clause and the absence of natural trigger synonyms a user would say.

Suggestions

Add a 'Use when...' trigger clause, e.g. 'Use when the user wants a web UI built or restyled from a mockup, screenshot, or live URL, or asks for pixel-perfect matching of a visual reference.'

Include natural user synonyms such as mockup, screenshot, restyle, and clone alongside 'generated references, static references, or live URL targets' so trigger-term coverage reaches the phrases users actually say.

Briefly name the boundary with adjacent skills (e.g. '$design briefs, $web-clone') in the description to sharpen distinctiveness rather than relying on the body for the distinction.

Dimension	Reasoning	Score
Specificity	Names concrete inputs ('generated references, static references, or live URL targets'), tooling ('$ultragoal with built-in visual verdict and pixel-diff evidence'), and a concrete end state ('leaves a reproducible design system'). It stops short of anchor 5 because it describes a pipeline at a high level rather than enumerating multiple concrete operations, and how the verdict/diff run is unstated.	4 / 5
Completeness	The 'what' is clear (orchestrates frontend implementation from a visual reference through verdict/pixel-diff iteration to a reusable design system), but there is no 'Use when...' clause or equivalent explicit trigger guidance; 'from generated references, static references, or live URL targets' only weakly implies when. Per the judging guideline, a missing 'Use when' clause caps completeness at 3.	3 / 5
Trigger Term Quality	Includes some relevant keywords users would say ('frontend UI', 'design system', 'live URL') but is missing common variations like mockup, screenshot, restyle, clone, or pixel-perfect, and leans on internal jargon ('Visual Ralph orchestration'). Not 4: key natural phrases users would actually use when needing this skill are absent.	3 / 5
Distinctiveness Conflict Risk	The niche (measured visual implementation loop with verdict evidence and design-system output) is mostly distinct from generic frontend or design skills, with 'Visual Ralph' branding adding distinctiveness. Not 5: it overlaps with adjacent design/mockup/frontend skill territory without naming the boundary in the description itself.	4 / 5
	Total	14 / 20 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
relative_links	Relative link issues: 1 suspicious	Warning

	Total	15 / 16 Passed

Repository: Yeachan-Heo/oh-my-codex
Path: skills/visual-ralph/SKILL.md
Commit: 1dcf513

Reviewed: 4 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.