span-of-control

Assessing Claude's span before accepting complex tasks prevents quality degradation in parallel execution. This skill evaluates parallel task count, task complexity, and coordination load to provide visible guardrails and explicit routing options.

When coordination load limits are exceeded during multi-tool orchestration, the skill provides structured options like sequential execution, scoped parallelism, or full parallel execution with quality risk notes. It requires explicit user decisions to proceed safely.

Defining a task contract for multi-task coordination involves scoping parallel tasks and verifying limits before starting. This skill integrates with structuring skills like contract-first-decomposition to extend span through better scoping and explicit user decisions.

Yes, this span assessment approach is suitable for long-running agentic workflows and multi-tool orchestration across sessions. It enforces guardrails by requiring structured task contracts when complexity or parallel task counts exceed reliable limits.

The best way to manage parallel task risk is to enforce span assessment across task complexity and coordination load before starting. This provides structured routing options like sequential execution or scoped parallelism to maintain output quality.

Claude produces inconsistent results on high-complexity tasks because it can accept work without visible operating span limits. Declaring reliable span before starting enforces guardrails and requires explicit user decisions to prevent unnoticed quality risks.