agentleFS
rahmanef63rahmanef63/si-coder-agent/skills/sc-fe/SKILL.md

sc-fe

Skill13 starsChanged 2 months agoMITView on GitHub

SI-Coder frontend quality orchestrator. Audit, design, implement, and verify frontend work across UI, UX, DX, and Agent Experience (AX), with reusable design presets/profiles such as --apple or --workbench and automatic preservation of existing design DNA by default.

The file

We split this skill into the rules it gives an agent

11 rules. About 0 are common, 7 are distinctive, and 4 are very specific to this project.

  • ≈0

    Common · estimate

    Many files like this have a rule of this kind.

  • ≈7

    Distinctive · estimate

    This author's own choices. Worth a look.

  • 4

    Very specific

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

  • Common
  • Distinctive
  • Very specific
  1. Distinctive:

    Keep durable instructions in English.

    line 15
  2. Distinctive:

    If multiple references conflict, prefer the more explicit user instruction, then product function, then accessibility/usability, then …

    line 67
  3. Very specific:

    Read references/profiles.md for preset meanings and the reusable profile contract.

    line 69
  4. Distinctive:

    When a project already has UI/UX and the user has not asked for a new direction:

    line 72
  5. Very specific:

    Treat user exclusions as hard boundaries.

    line 87
  6. Distinctive:

    Read project structure, design system, current behavior, and rendered UI where possible.

    line 93
  7. Distinctive:

    test key interactions and keyboard path,

    line 112
  8. Very specific:

    verify loading/empty/error states,

    line 113
  9. Distinctive:

    verify responsive resize behavior,

    line 115
  10. Distinctive:

    verify AX/DX contracts when agent/developer surfaces changed.

    line 116
  11. Very specific:

    When sc-all is building/publishing a user-facing app, sc-all should delegate frontend quality to sc-fe before final production verification.

    line 125

≈ “Common” is an estimate: at least 1 in 5 of the 350,495 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

sc-fe — frontend quality orchestrator

Language

Keep durable instructions in English. Reply in the user's language unless requested otherwise.

Ownership

sc-fe composes four quality axes:

  • sc-ui — visual hierarchy/craft/design system/anti-slop,
  • sc-ux — human usability/accessibility/interaction/recovery,
  • sc-dx — developer ergonomics/maintainability/debug/verification,
  • sc-ax — Agent Experience/discoverability/operability/resume/context efficiency.

Normal users do not need to choose a sub-skill. Route internally. Direct sub-skill invocation remains valid for focused audits.

Invocation and flags

Host syntax varies, but on slash-capable hosts use patterns such as:

/sc-fe improve the desktop shell
/sc-fe --apple improve settings
/sc-fe --workbench audit the desktop shell
/sc-fe --profile baton-desktop refine the inspector
/sc-fe --save-profile baton-desktop
/sc-fe --apple --density compact --motion subtle
/sc-fe --audit --strict

Supported semantic flags:

  • --apple → shorthand for --preset apple.
  • --workbench, --linear, --notion, --vercel, --material, --editorial, --terminal → shorthand named presets.
  • --preset <name> → built-in principle preset.
  • --profile <name> → load a project-defined reusable frontend profile.
  • --save-profile <name> → extract current design DNA and save a reusable project profile when filesystem write access exists.
  • --existing → explicitly preserve/extend existing design DNA. This is already the default for an existing coherent product.
  • --fresh → allow a new design direction when no existing profile should constrain it.
  • --density compact|comfortable|spacious.
  • --motion none|subtle|expressive.
  • --platform desktop|mobile|responsive.
  • --audit → report first; do not make cosmetic invention merely to create work.
  • --strict → treat UI/UX/DX/AX verification failures as completion blockers when applicable.
  • --technical → expose implementation details normally kept internal.

Unknown flags must not be silently invented. Explain unsupported input or treat it as plain-language intent.

Profile resolution order

Resolve the frontend language in this order:

  1. explicit user scope/exclusions and functional requirements,
  2. explicit flag overrides (--density, --motion, platform, etc.),
  3. explicit named project --profile,
  4. explicit built-in --preset / shorthand such as --apple,
  5. coherent existing project design DNA,
  6. product/platform defaults.

An explicit preset does not erase functional requirements or accessibility. An existing coherent product wins over generic SI-Coder taste unless the user asks to redesign it.

If multiple references conflict, prefer the more explicit user instruction, then product function, then accessibility/usability, then aesthetic preference.

Read references/profiles.md for preset meanings and the reusable profile contract.

Existing UI learning

When a project already has UI/UX and the user has not asked for a new direction:

  1. inspect rendered screens when possible,
  2. inspect tokens/theme/components/layout primitives,
  3. identify repeated visual and interaction rules,
  4. distinguish intentional system from accidental inconsistency,
  5. preserve the intentional rules,
  6. fix inconsistencies against that system instead of importing a new template.

--save-profile <name> stores the extracted contract at .sc/frontend/profiles/<name>.json when project writes are allowed. Never store screenshots, copyrighted assets, secrets, or huge generated dumps in the profile.

For repeated-control inconsistencies, workbench responsiveness or Svelte runtime migrations, read shared frontend contracts. Resolve registry, component, state and token ownership before editing individual screens.

Scope lock

Treat user exclusions as hard boundaries. Example: if the user says mobile nav dock jangan disentuh, do not edit it, its styles, its behavior, or shared abstractions in a way that changes it. If a required shared change risks the excluded surface, isolate the implementation or report the conflict.

Workflow

For substantial frontend work:

1. Inspect

Read project structure, design system, current behavior, and rendered UI where possible. Do not redesign from filenames alone.

2. Design Read

Internally summarize product type, user jobs, platform, density, hierarchy, brand/design DNA, state model, constraints, and profile/preset resolution.

3. Audit four axes

Score/prioritize defects by impact, not by how easy they are to make pretty:

  • UI: hierarchy/craft/consistency/anti-slop,
  • UX: task flow/feedback/accessibility/recovery,
  • DX: code/system/debug/change friction,
  • AX: agent orientation/tool semantics/resume/context waste.

4. Implement smallest coherent set

Fix root system issues before local decoration. Reuse current components/tokens. Avoid broad rewrites when targeted changes solve the actual problem.

5. Verify

Frontend completion requires more than compilation. When tools allow:

  • format/lint/typecheck/tests/build,
  • inspect rendered pixels at representative viewports,
  • test key interactions and keyboard path,
  • verify loading/empty/error states,
  • confirm no excluded surface changed,
  • verify responsive resize behavior,
  • verify AX/DX contracts when agent/developer surfaces changed.

6. Report evidence

Report problem → change → verification → remaining risk. Do not claim subjective quality as verified fact.

Desktop shell default

For application shells, developer tools, admin workbenches, and agent consoles, default to a workbench mental model rather than a marketing dashboard: clear frame/navigation/work-area/contextual-tools regions, dense-but-readable controls, keyboard-first acceleration, visible selection/focus/status, deterministic pane/resize behavior, and restrained decoration.

Integration with sc-all

When sc-all is building/publishing a user-facing app, sc-all should delegate frontend quality to sc-fe before final production verification. Backend-only or infrastructure-only tasks do not need this gate. Forward any frontend flags/preset/profile named by the user.

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.