agentleFS
ytvee-devytvee-dev/webdev-agent-kit/skills/frontend-refactor-surgeon/SKILL.md

frontend-refactor-surgeon

Skill7 starsChanged 21 days agoApache-2.0View on GitHub

Perform bounded, behavior-preserving frontend refactors such as component extraction, file moves, prop or interface cleanup, state simplification, and TypeScript tightening. Excludes bugfixes, features, redesign, packages, and unapproved behavior changes.

The file

Show its rules and text

We split this skill into the rules it gives an agent

22 rules. About 0 are common, 10 are distinctive, and 12 are very specific to this project.

  • ≈0

    Common · estimate

    Many files like this have a rule of this kind.

  • ≈10

    Distinctive · estimate

    This author's own choices. Worth a look.

  • 12

    Very specific

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

  • Common
  • Distinctive
  • Very specific
  1. Distinctive:

    Use this skill when the user asks to simplify or reorganize frontend code, extract components, hooks, utilities, or styles, tighten …

    line 46
  2. Distinctive:

    Do not use this skill for bugfix-first debugging, new feature implementation, design/spec work, visual QA only, broad rewrites without a …

    line 50
  3. Very specific:

    Use frontend-bugfix-debugger when the primary goal is to fix a defect.

    line 52
  4. Very specific:

    Read common/prompt-intent-routing-rules.md when workflow level is unclear.

    line 57
  5. Very specific:

    Read common/approved-patterns.md for component decomposition rules.

    line 59
  6. Very specific:

    Read common/anti-patterns.md and relevant anti-pattern templates when the refactor touches components.

    line 60
  7. Very specific:

    Read common/verification-loop-rules.md and common/bounded-retry-rules.md when verification repair is in scope.

    line 61
  8. Very specific:

    Read relevant boundary docs such as common/state-ownership-rules.md when the refactor touches state.

    line 63
  9. Distinctive:

    Read project overlays and affected source files needed to define the behavior boundary.

    line 64
  10. Very specific:

    When shared UI or compatible component wrappers are in scope, read common/frontend-design-system-rules.md and …

    line 66
  11. Very specific:

    When React/Next work changes loading, imports, subscriptions or expensive rendering, read the relevant section of …

    line 71
  12. Distinctive:

    Must not install packages, add testing workflows, change build tooling, add UI libraries, or migrate frameworks without explicit approval.

    line 80
  13. Distinctive:

    Must not change production systems, secrets, or production data.

    line 81
  14. Distinctive:

    Do not create or preserve components that mix routing, data access, state orchestration, transformations, form logic, repeated markup, …

    line 91
  15. Very specific:

    Do not hide missing decomposition behind renderXxx, xxxRender, nested array pipelines, component-body JSX preparation, oversized custom …

    line 92
  16. Distinctive:

    Run the smallest relevant verification after meaningful edits.

    line 93
  17. Distinctive:

    If verification fails because of the current refactor, fix related failures using bounded retry rules and keep the behavior contract …

    line 94
  18. Very specific:

    Use frontend-linter-manager after code-changing work when a lint command exists.

    line 95
  19. Very specific:

    Use frontend-visual-qa when rendered UI could have changed.

    line 96
  20. Very specific:

    Use frontend-quality-reviewer when the refactor is standard or deep, involved retry, or the user asks for review.

    line 97
  21. Distinctive:

    Stop and ask before behavior change, architecture expansion, dependency changes, or broad rewrite.

    line 98
  22. Distinctive:

    No testing workflow, UI library, package install, or framework migration may be introduced by default.

    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

Frontend Refactor Surgeon

Purpose

Refactor frontend code while preserving behavior, rendered output, and public contracts unless the user explicitly approves a behavior change.

Refactor verification may use a bounded loop for related failures, but the refactor must not expand into feature work or broad rewrite.

When To Use

Use this skill when the user asks to simplify or reorganize frontend code, extract components, hooks, utilities, or styles, tighten TypeScript types, remove duplication within a clear frontend boundary, decompose mixed-responsibility components, or prepare code for a future change without adding the feature yet.

When Not To Use

Do not use this skill for bugfix-first debugging, new feature implementation, design/spec work, visual QA only, broad rewrites without a boundary, backend refactors, package installation, or testing workflow creation.

Use frontend-bugfix-debugger when the primary goal is to fix a defect.

Required Context

  1. Read AGENTS.md.
  2. Read common/prompt-intent-routing-rules.md when workflow level is unclear.
  3. Read common/refactor-safety-rules.md.
  4. Read common/approved-patterns.md for component decomposition rules.
  5. Read common/anti-patterns.md and relevant anti-pattern templates when the refactor touches components.
  6. Read common/verification-loop-rules.md and common/bounded-retry-rules.md when verification repair is in scope.
  7. Read common/typescript-discipline.md.
  8. Read relevant boundary docs such as common/state-ownership-rules.md when the refactor touches state.
  9. Read project overlays and affected source files needed to define the behavior boundary.

When shared UI or compatible component wrappers are in scope, read common/frontend-design-system-rules.md and common/component-substitution-rules.md. Apply purpose-specific naming from common/frontend-implementation-boundaries.md to changed project-owned code.

When React/Next work changes loading, imports, subscriptions or expensive rendering, read the relevant section of common/react-performance-rules.md.

Tool Contract

  • May inspect affected source, styles, configs, and project overlays.
  • May run existing lint, typecheck, build, or preview commands relevant to the refactor.
  • May use Browser or Playwright MCP when rendered output must be checked.
  • Activate openai_platform_docs only when current OpenAI API or ChatGPT Apps SDK behavior affects the preserved contract.
  • Must not install packages, add testing workflows, change build tooling, add UI libraries, or migrate frameworks without explicit approval.
  • Must not change production systems, secrets, or production data.

Workflow

  1. Define the refactor boundary and behavior contract.
  2. Identify public APIs, rendered output, route behavior, state ownership, accessibility, and TypeScript contracts that must remain stable.
  3. Identify whether the refactor requires decomposition.
  4. Split the refactor into small mechanical steps.
  5. Edit one boundary at a time.
  6. Preserve project naming, folder, styling, and state conventions.
  7. Do not create or preserve components that mix routing, data access, state orchestration, transformations, form logic, repeated markup, large JSX, and side effects in one file.
  8. Do not hide missing decomposition behind renderXxx, xxxRender, nested array pipelines, component-body JSX preparation, oversized custom hooks, or unnecessary useCallback.
  9. Run the smallest relevant verification after meaningful edits.
  10. If verification fails because of the current refactor, fix related failures using bounded retry rules and keep the behavior contract unchanged.
  11. Use frontend-linter-manager after code-changing work when a lint command exists.
  12. Use frontend-visual-qa when rendered UI could have changed.
  13. Use frontend-quality-reviewer when the refactor is standard or deep, involved retry, or the user asks for review.
  14. Stop and ask before behavior change, architecture expansion, dependency changes, or broad rewrite.

Output Contract

Final response: return only facts that affect the user's understanding, confidence, or next action. Omit empty fields and workflow narration.

Refactor boundary:
Behavior preserved:
Decomposition applied:
Changes made:
Attempts:
Verification:
Rendered QA:
Deferred work:
Risks:

Validation Gates

  • The behavior boundary must be explicit before editing.
  • Public contracts must remain stable unless approved.
  • Refactor must not introduce feature work by default.
  • Verification repair must not change behavior to make checks pass.
  • Component decomposition rules must be applied regardless of framework, router, state layer, data layer, or styling system.
  • Lint verification must run after code changes when a command exists.
  • Rendered UI changes or risk must trigger visual QA when applicable.
  • No testing workflow, UI library, package install, or framework migration may be introduced by default.

Trigger Evals

Should trigger:

  • "Extract this component while preserving its public behavior."
  • "Simplify this state flow in bounded verified steps."

Should not trigger:

  • "Fix this runtime bug from its reproduction."
  • "Redesign this page."

Reference Map

  • common/react-performance-rules.md
  • common/refactor-safety-rules.md
  • common/approved-patterns.md
  • common/anti-patterns.md
  • common/verification-loop-rules.md
  • common/bounded-retry-rules.md
  • common/typescript-discipline.md
  • common/state-ownership-rules.md
  • project/verification-profile.md

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.