palarran-frontend-build

原文涵盖的主题

• Frontend build
• Mandatory guides and memory
• SDD and actual execution
• Browser gates, DoD and progress

Shared cycle: before work and after checks, perform global memory: bounded recall → applicability → sanitized verified learn/dedup. This memory scope is explicitly authorized; other profile prohibitions remain. --plan uses no network/publication; an outage does not block the primary task. Invoke /palarran:frontend:build {plan path/ID or scope/UX/UI paths} or $palarran-frontend-build. Work in the target application using its existing stack; authoring this skill does not imply building a palarran frontend. Separate the skill checkout, found from the real SKILL.md with symlinks resolved (../../), from the target project. Read decision memory, target shared/UX/UI/frontend indexes, relevant accepted ADR/spec/plan/handoffs and prior progress. Explain every accepted implementation choice before dependent changes: basis, alternatives, evidence, consequences; preserve canonical ADR/history/index/handoff/reuses/supersedes for later tasks. Old authorization does not grant new scope. Required reading: Contract, source UX/UI/SDD mapping. SDD/agent worktrees/DoD, frontend progress, shared execution protocol. Implementation/data/assets, responsiveness. Browser acceptance, quality. Workflow/contex

这里只展示节选，请以完整原文和当前要求为准。