tdd-feature-agent

Orchestrating feature development across multiple agents requires central coordination to read progress, delegate tasks to a coder, and manage handoffs. This ensures focused execution and clean context isolation for each sub-agent.

Automating code review after implementation involves delegating the completed feature to a code-review agent. The orchestrator initiates this review immediately after the coder agent finishes the assigned feature work.

Handling test failures in automated workflows requires routing the failed execution to a debugger or deep-dive agent. These specialized agents perform root-cause analysis and remediation to resolve the underlying issues.

Isolated context windows are used when delegating coding tasks to maintain safe handoffs and task isolation. Spinning up fresh contexts prevents cross-contamination of instructions and ensures each agent focuses strictly on its specific objective.

Tracking pending feature dependencies using a JSON file provides the central context needed for agent orchestration. The coordinator identifies the next pending feature and its dependencies directly from this file to initiate implementation.

Agent orchestration works for recovering from debugging failures by automatically engaging deep-dive agents. These agents analyze root causes of failed tests and update progress tracking accordingly to resume the workflow.