harness:dispatching-parallel-agents

What problem does it solve?

It reduces wasted time by helping you investigate multiple independent failures concurrently instead of serially, so each agent stays focused on one problem domain.

Core Features & Use Cases

• Parallel agent dispatch: Breaks work into independent scopes and assigns one agent per domain to avoid context bleeding and interference.
• Guardrails for safe parallelization: Enforces the harness local rule to read harness-delta.md and to block orphan parallel work mapped to features.json.
• Integration-ready workflow: Reviews each agent’s findings, checks for conflicts, and recommends running the full test suite to verify all fixes together.

Use case: When 3 different test files fail with unrelated root causes (e.g., abort behavior, batch completion, and tool-approval race conditions), dispatch three agents so they can fix each domain simultaneously and then integrate results.

Quick Start

Use the harness:dispatching-parallel-agents skill when you have 2+ independent tasks and want each agent to work on a single scoped failure domain in parallel.