stitch-loop

Iterative design-to-code feedback loop. Critique → adjust → ship cycle for tightening visual fidelity between brief and built UI.

Quality

46%

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/stitch-loop/SKILL.md

Quality

Content

43%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.

This is a catalog stub, not a working skill: the body is lean and cleanly sectioned, but it contains essentially no executable guidance — the actual critique → adjust → ship workflow is entirely absent, outsourced to a browser-opened upstream README the agent cannot use. The simple-skill exceptions in the rubric cannot apply because the single advertised action is not unambiguous.

Suggestions

Inline the core loop as a numbered sequence with concrete steps (compare built UI against the brief, enumerate visual discrepancies, apply fixes, re-render/review) so the skill works without fetching the upstream repo.

Replace the browser-oriented `open https://...` command with an agent-executable instruction (e.g., a `git clone` or fetch of the upstream bundle and the exact path to install it into), or bundle the needed reference files locally under references/.

Add validation checkpoints to the loop (e.g., "stop iterating when the rendered output matches the brief within tolerance") since the entire point of the skill is an iterate-until-good cycle.

Dimension	Reasoning	Score
Conciseness	The body is short and unpadded with no explanation of concepts Claude already knows; every section is brief. Minor trimming is possible — the description is repeated in "What it does" and the upstream URL appears in both the Source section and the code block — which keeps it below score 5.	4 / 5
Actionability	The only commands are `open https://github.com/google-labs-code/skills` (a browser open, not usable by a coding agent) and a vague instruction to "install the upstream bundle into your active agent's skills directory" with no install command. The advertised critique → adjust → ship workflow has no concrete executable steps at all. Not score 1 because there is at least one literal command and a specific repo URL.	2 / 5
Workflow Clarity	The multi-step process the skill is named for ("Critique → adjust → ship cycle") is only named, never sequenced — there are no critique criteria, adjustment steps, or ship/verify checkpoints anywhere in the body. The "How to use" section has a rough sequence (inspect upstream, install, invoke) with major gaps and no validation. Not score 1 because a rough install-and-invoke sequence is present.	2 / 5
Progressive Disclosure	Sections (What it does / Source / How to use) are clearly organized and the upstream reference is clearly signaled, but no local bundle files exist (references/, scripts/, assets/ are absent), so the actual reference material sits behind an external GitHub URL rather than one-level-deep bundled files. The skill is short enough that structure is not the failure, but the pointer-only organization leaves discovery of the real content to the agent.	3 / 5
	Total	11 / 20 Passed

Description

50%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 names a clear niche and a recognizable action cycle, but reads more like a tagline than a capability description: no "when to use" guidance, thin action coverage, and triggers that are half natural phrases, half project-specific jargon. It sits at the midpoint of the rubric.

Suggestions

Add an explicit "Use when..." clause naming the situations that call for this skill (e.g., "Use when the built UI visually diverges from the design brief or mockup and needs iterative tightening").

Enumerate the concrete actions the loop performs (e.g., compare rendered UI against the brief, list visual discrepancies, apply code fixes, re-render) instead of only naming the abstract cycle.

Add natural trigger synonyms users would actually say ("UI doesn't match the design", "design QA", "mockup to code") and drop or de-emphasize the project-specific "stitch loop" and "fidelity loop" phrasing.

Dimension	Reasoning	Score
Specificity	Names the domain ("design-to-code feedback loop") and 1-2 actions ("Critique → adjust → ship cycle"), but never states what artifacts or operations are involved, so coverage is not comprehensive. It is not score 2 because concrete actions are named alongside the domain, and not score 4 because there are not several specific actions with only minor gaps.	3 / 5
Completeness	A clear "what" is present ("Iterative design-to-code feedback loop... tightening visual fidelity"), but there is no "Use when..." clause or equivalent explicit trigger guidance in the description itself, which per the judging guidelines caps completeness at 3. Not score 2 because the "what" is clear, not vague.	3 / 5
Trigger Term Quality	The frontmatter triggers include some natural phrases ("design to code", "design iteration") but miss common variations users would say ("UI iteration", "design QA", "visual review", "mockup to code"), and "stitch loop" / "fidelity loop" are niche jargon. Not score 4 because a few natural terms are clearly missing.	3 / 5
Distinctiveness Conflict Risk	The niche of visual fidelity between brief and built UI is somewhat specific, but broad trigger phrases like "design to code" and "design iteration" would overlap with general UI-generation, frontend-development, and design-system skills. Not score 4 because the overlap risk with closely related skills is more than minor.	3 / 5
	Total	12 / 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
frontmatter_unknown_keys	Unknown frontmatter key(s) found; consider removing or moving to metadata	Warning

	Total	15 / 16 Passed

Repository: nexu-io/open-design
Path: skills/stitch-loop/SKILL.md
Commit: 1b47e60

Reviewed: 3 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.