manager-with-local

Persist task delegation chains for crash recovery by writing local breadcrumbs before each manager call, logging completed steps with files_changed, and explicitly completing or aborting to avoid orphaned operations.

Local breadcrumb lifecycle management for multi-step task orchestration is the process of starting a breadcrumb, logging each completed delegation step with files_changed, and explicitly completing or aborting to ensure auditable, crash-recoverable tracking.

Yes, you need both manager and local servers installed for delegation tracking. This Skill applies when you have coordinated, multi-step delegations or loaf-driven task groups that may exceed a few minutes and require durable cross-context persistence.

Resume a multi-step refactor after an agent context reset by reading the persisted breadcrumb status. A new session or agent reads the local breadcrumb logs and resumes the operation at the correct next step with full audit context.

Handle rollback and retry patterns in long-running pipelines by recording failed steps in local breadcrumbs, logging failure context for safe decisions, and executing explicit retry or rollback operations to maintain auditable operation tracking.

You should not use local breadcrumbs for task orchestration if your delegations are short-lived, single-step operations, or if you lack the required manager and local servers, as the breadcrumb lifecycle overhead provides no crash recovery benefit for ephemeral tasks.