orchestration

Routing tasks across multiple AI providers involves using centralized decision logic to select the appropriate provider based on task characteristics, then delegating execution. This skill coordinates that routing, allowing you to direct coding tasks to OpenAI Codex, Google Gemini, or other agents and consolidate their results into a single output.

Parallel execution with fallback chains lets you launch multiple providers simultaneously and automatically switch to alternatives when one fails. This skill enables that pattern, ensuring task completion even when a primary provider is unavailable or underperforms.

Aggregating results from heterogeneous agents requires clean result collection with consistent formatting across providers. This skill provides patterns for native-invoke delegation and result merging, so you get unified output from parallel or sequential agent runs without manual consolidation.

Delegation between execution environments means routing a task from one agent or sandbox to another and retrieving the result. This skill handles non-interactive orchestration across environments, enabling seamless handoffs without manual intervention or result parsing.

Orchestration works across heterogeneous AI providers and execution environments, not just a single vendor. This skill applies to multi-provider scenarios where you need consistent coordination logic regardless of whether providers are cloud-based, sandboxed, or specialized agents.

Provider selection logic matters when different tasks have different performance or capability requirements. Use it when you need to route code reviews to one provider, fast validation checks to another, and need automated decision-making rather than manual provider assignment for each task.