What problem does it solve? Coordinating multiple AI workers, dependency graphs, and scheduled tasks in Jcode is error-prone: lost worker IDs, premature success claims, duplicate spawns after timeouts, and unsafe cleanup. This Skill defines a disciplined orchestration workflow so delegated work is bounded, verified, and recoverable. ## Core Features & Use Cases - Bounded delegation: Spawn workers with explicit contracts covering acceptance checks, file ownership, report format, and stop conditions. - Dependency graphs and waits: Build task graphs with depends_on edges, await members by exact session IDs, and verify settled state before advancing dependents. - Durable goals and schedules: Track initiatives, milestones, and future schedules with checkpoints that survive session restarts. - Recovery and cleanup: Reconcile live state after interruptions, retry only understood failures, and cancel or stop only owned workers. - Use Case: You need a codebase review, a test run, and a documentation update executed in parallel. Spawn three workers with distinct write ownership, await their completion, verify each report against actual artifacts and test results, then checkpoint the milestone. ## Quick Start Ask the agent to spawn parallel Jcode workers for your subtasks, wait for their completion, and verify each report before marking the work done.
