claudish-usage

Sub-agent delegation routes external model prompts through isolated contexts, preventing main-context pollution. You create file-based instruction patterns to run tasks with chosen models, keeping the primary conversation clean and reproducible.

Workflow orchestration with external models uses sub-agent chains and file-based instruction patterns. This approach isolates model-specific prompts and routing logic, allowing you to manage complex multi-agent tasks reproducibly.

Yes, you can run Claude Code with OpenRouter models by routing requests through sub-agents. This integration isolates external model interactions from the main context while enforcing controlled delegation patterns.

Reproducible workflows use file-based instruction patterns to define sub-agent tasks, model selection, and routing. You write instructions to a file, execute the sub-agent chain, and read results from the generated output file.

Sub-agents provide context isolation, preventing external model prompts from leaking into the primary context. This ensures controlled delegation, maintains main conversation clarity, and enforces structured workflow orchestration for complex tasks.

Sub-agent delegation supports model selection and routing guidance for external AI models. You can specify different models for different sub-agents within your file-based instruction pattern to match task requirements.