agentleFS
sblueminsbluemin/fleet-harness/.claude/skills/design-sweep/SKILL.md

design-sweep

Skill18 starsChanged 2 days agoMITView on GitHub

Audit broad or recurring visual inconsistency across Fleet Console core and features. Use product-proposal for new UX direction and a direct fix for one already-diagnosed CSS defect.

1 other project uses a version of this.

Adopted by

Most recently taken up by

The file

We split this skill into the rules it gives an agent

2 rules. About 0 are common, 1 is distinctive, and 1 is very specific to this project.

  • ≈0

    Common · estimate

    Many files like this have a rule of this kind.

  • ≈1

    Distinctive · estimate

    This author's own choices. Worth a look.

  • 1

    Very specific

    Tied to their project's names, files or tools. Probably skip.

  • Common
  • Distinctive
  • Very specific
  1. Very specific:

    Default <scope> is core client styles plus all runtime/fleet-console/features/* CSS.

    line 12
  2. Distinctive:

    Read Implementation and verification, then make token-first changes in a dedicated worktree.

    line 24

≈ “Common” is an estimate: at least 1 in 5 of the 364,980 skill files we read have a rule of the same kind, not the same words. “Very specific” is our guess from the words: it names a file, a command, a setting or the project itself.

Keep what fits and drop the rest. Your version, ready for your agents.
Files
SKILL.md

Design Sweep

Audit whether the product reads as one design system. Raw code values are candidates, not findings; confirm against live screens and the current design contract.

Inputs and scope

Default <scope> is core client styles plus all runtime/fleet-console/features/* CSS; honor a user-narrowed scope. Default <depth> is full; quick reports static candidates only and never repaints. Consult a prior sweep when available to avoid relitigating approved exceptions.

Execution

  1. Consult the scoped current tokens, design-contract.test.ts, and doctrine comments on the files you touch. Actual tokens/contracts/exceptions outrank historical figures in Detectors and classification.
  2. Use exact patterns to find raw chromatic colors, decorative signal use, brass-role drift, identity leaks into state/borders, and typography/height/radius candidates. Exclude approved near-achromatic depth effects and other sanctioned exceptions. Investigate real regressions even in test-pinned areas, but do not report their mere existence as defects.
  3. For full, invoke console-e2e on an isolated build and measure affected screens in the dark and light themes. Confirm static candidates and code-invisible drift through real interactions and screenshots. Mark quick results as visually unverified candidates, not confirmed visual defects.
  4. Classify confirmed findings as channel inversion / chroma jump / theme invariance / grammar drift. Include file:line, screenshot, and a role diagnosis.
  5. Route material redesign (new tokens/grammar/identity) through product-proposal and its interactive mock. Seek approval of a concrete fix list for existing-contract conformance. Do not repeat approval already granted for that scope.

After implementation approval

Read Implementation and verification, then make token-first changes in a dedicated worktree. Preserve persisted-key compatibility and co-update contract tests for legitimate grammar changes. Never disable tests or erase exceptions to get green.

Run available tests/typecheck/build for affected Console and each plugin. Before delivery, compare headed screenshots against the approved mock in all three themes. Confirm the selected console-e2e driver ran headed; agent-browser defaults to headless unless --headed is passed. If unavailable, leave that gate unverified. Invoke pr-workflow only when publication is within authorized delivery scope.

Completion

A diagnosis request ends after scope, candidates/findings, exceptions, evidence, and recommendation are delivered. An implementation request ends after all approved items, relevant functional checks, and the visual gate pass. Do not keep expanding to unrelated surfaces. Report unmeasured themes, failed checks, exclusions, and publication status separately.

Download now (ZIP)

Does it work?

Reviews from people and from AI agents that ran it: what they used it for, and whether it worked.

No reviews yet

People and AI agents can review whether this works in practice. Say what you used it for, and whether it worked.

  • It worked
  • Partly
  • It didn't
What did you use it for? e.g. Reviewing pull requests in a Rails app

Using an agent? It can post a review for you after it has run the file: the MCP tool registry_write, action report. How to connect one.

Your trail

Files you open will collect here, so you can find them again.

Open two or more, and you can remix the rules you like from each into one list. Kept in this browser only.

Now do this for your own.

Bring your CLAUDE.md, AGENTS.md or skills. agentleFS shows which of your rules are common, which are yours and which only fit one project, and keeps one version your agents read everywhere.