webdev-agent-kit

Review frontend quality from evidence across code, UI, architecture, TypeScript, security, performance, verification, decomposition, UX, and anti-slop concerns. Returns pass, concerns, or fail without implementing unrequested fixes.

Define a platform-neutral bounded loop for standard or deep work using measurable acceptance criteria, verification, retry limits, independent review, memory, and stop conditions. Skip lookups, tiny edits, and one-off explanations.

Create, resume, analyze, or converge standard and deep frontend execution plans with traceable slices, evidence, and stop/resume state. Skip lightweight isolated edits.

Define user goal, scope, constraints, and done criteria before standard or deep frontend work. Skip micro-fixes, obvious type errors, tiny styling changes, and isolated direct edits.

Create, evaluate, or edit this .agents WebDev Agent Kit, including AGENTS.md, common/**, project/**, skills/**, metadata, validators, and progressive-disclosure rules.

Plan architecture and ownership for standard or deep React/Next.js work across routing, state, data, styling, forms, builds, migrations, and implementation handoff. Skip micro-fixes and unrelated stacks.

Diagnose and fix frontend defects from evidence, including unclear UI/runtime errors, broken routes, styling regressions, and hydration or client issues. Reproduce before editing; exclude planned refactors and backend-only debugging.

Implement an approved Design Implementation Spec or screenshot-derived spec in React/Next.js using existing CSS Modules, Redux, TanStack, and Axios conventions. Excludes unrelated stacks, live design tools, and new styling systems.

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.

Plan a new frontend project or first vertical slice from a product idea, with approval gates before scaffolding. Excludes existing-project micro-fixes, automatic scaffolds, package installation, UI libraries, and test setup.

Detect frontend tool capabilities, report gaps, verify official setup sources, plan approval-gated MCP changes, or update project/mcp-profile.md. Prefer capabilities over provider names; never install or configure without approval.

Add or tighten reusable approved patterns and anti-patterns in common/** with source-backed examples, approvals, graph links, evals, and validation. Keep project-specific patterns in local-only project/**.

Refresh local-only project/** facts and frontend path indexes after stack, routing, styling, asset, verification, docs/tool, design-reference, pattern, or ownership changes. Do not put project facts in reusable skills.

Adapt this .agents bundle to a frontend project using approved native pointers, local-only overlays, stack detection, skill applicability, tool capabilities, client and verification facts, optional loop memory, and context cache. Do not create app code or overwrite instructions.

Turn user-supplied screenshots, copied inspect values, exports, assets, or design notes into a strict frontend Design Implementation Spec. Live design links alone are insufficient.

Set subject-grounded visual direction for standard or deep UI work, redesign, polish, critique, anti-template review, or design handoff. Skip technical micro-fixes and isolated code edits.

Ground standard or deep UI work in product category, page pattern, design dials, domain UX risks, and anti-patterns before visual direction. Produces a brief, not code or tool setup.

Run scoped frontend lint verification, repair related lint failures, or plan explicitly requested lint setup. Prefer existing commands; dependencies, scripts, and config changes require approval.

Verify rendered frontend evidence after implementation, including screenshot comparison, responsiveness, overflow, visible states, and scoped visual regression. Use available browser tooling only; exclude static lookup, routine code review, lint, builds, onboarding, and tool detection.