palarran-backend-plan

原文涵盖的主题

• Backend plan
• Boundaries and memory
• Mandatory foundation
• Workflow
• Future implementation progress

Shared cycle: before work and after checks, follow global memory: bounded recall → applicability → sanitized verified learning/dedup. This memory scope is explicitly authorized; other profile restrictions remain. --plan has no networking/publication; outages do not block the primary task. Invoke /palarran:backend:plan {feature} or $palarran-backend-plan {feature}. Planning only. Authoring this skill does not mean planning Palarran's backend. Locate the installed checkout from the actual SKILL.md, resolving its symlink (../../), separately from the target project. Read shared memory, target .palarran/INDEX.md, backend index and relevant accepted ADRs/specs/handoffs. Explain each accepted choice before dependent planning changes: rationale/alternatives/evidence/consequences; update canonical decisions/index/handoff and reuses/supersedes. Blocked proposals are not accepted; old ADRs grant no fresh authority. Preserve later-task context in target, not plugin caches. Allowed: read-only local repository research, web/official-doc research, plan/spec/contracts/design sketches/diagrams and decision records in target planning areas or existing canonical spec conventions. Forbidden: applicat

这里只展示节选，请以完整原文和当前要求为准。