agentic-workflow

You orchestrate a multi-agent development pipeline by coordinating sequential stages: research, planning, validation, TDD implementation, and review. Agents execute in the background and transfer artifacts via file-based handoffs to keep the main context window clean and token-efficient.

File-based handoffs manage multi-agent coordination by saving intermediate data to `.claude/cache/agents/`. This ensures reliable data transfer between sequential development stages without consuming the main context window with background execution logs.

To enforce Test-Driven Development in automated pipelines, mandate a dedicated TDD implementation stage within your multi-agent workflow. This ensures agents write and validate tests alongside code generation before the final review stage verifies the artifacts.

Yes, you can manage agent context windows by utilizing `run_in_background: true` during pipeline execution. This keeps the main context clean and token-efficient while background agents process complex research, planning, and implementation tasks.

You need a structured multi-agent workflow for complex software development tasks requiring sequential collaboration and artifact verification, such as implementing a new feature that requires researching best practices, planning code structure, and validating the implementation.