palarran-backend-build

原文涵盖的主题

• Backend build
• Foundation, scope and memory
• Execution

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:build {plan path/ID} or $palarran-backend-build. Implement the target application from a ready source plan, not Palarran's backend while authoring the skill. Find the skill checkout from actual SKILL.md, resolving its symlink (../../), separately from target root. Read shared memory, target shared/backend indexes and relevant accepted ADRs/specs/plans/execution handoffs. Explain accepted implementation choices before dependent changes: rationale, alternatives, evidence, consequences; retain canonical ADR/history/reuses/supersedes/index in target. Old permission is not new scope. Mandatory reading: Build contract, source/progress/resume, shared progress contract. Agent worktrees/integration, implementation/data/sandbox. Edge/integration/DoD, SLA/SLO validation. Workflow, sources. Progress example, execution agent, verification, handoff, progress helper. Source methodology:

这里只展示节选，请以完整原文和当前要求为准。